Method, apparatus and computer readable medium for analysing the condition of a machine having a rotating part

The present invention relates to a method for analyzing a machine condition and to a device for analyzing a machine condition. The invention also relates to a system including this device, and to a method of working with such a device. The invention also relates to a computer program for instructing a computer to perform an analytical function.

State of the art

Machines with moving parts are subject to wear over time, which often leads to deterioration of the condition of the machine. Examples of such machines with moving parts are electric motors, pumps, generators, compressors, lathes and CNC machines. Moving parts may include shaft and bearings.

To prevent machine malfunction, such machines must undergo maintenance depending on the condition of the machine. Therefore, the operating condition of such a machine is preferably periodically evaluated. The operating state can be determined by measuring the vibrations emanating from the bearing, or by measuring the temperature on the machine body, these temperatures depending on the operating state of the bearing. Such condition checks of machines with rotating or other moving parts are of great importance for safety, as well as for the service life of such machines. A known method is the manual execution of such measurements on machines. This is usually done by the operator using a measuring device that takes measurements at measurement points on one or more machines.

A number of commercially available measuring instruments are available, which are based on the fact that defects in the roller bearings form short pulses, usually called shock pulses. The shock pulse measuring device may generate information indicative of the state of the bearing or machine.

\ νϋ 03062766 discloses a machine having a measuring point and a shaft with a specific shaft diameter, wherein the shaft can rotate when the machine is in use. νθ 03062766 also discloses a device for analyzing the condition of a machine having a rotating shaft. The disclosed device has a sensor for generating a measured value indicating vibration at a measurement point. The device disclosed in νθ 03062766 has a data processor and a storage device. The storage device may store program code, which, when executed in the data processor, instructs the analysis device to perform the function of monitoring the state of the machine. Such a machine condition monitoring function may include shock pulse measurement.

υδ 6053047 discloses an accelerometer used as a vibration sensor that collects analog vibration data delivered to an analog-to-digital converter that provides digital vibration data to the processor 90. According to υδ 6053047, the processor digitally filters the digital vibration data, rectifies the filtered signal and low pass filtering of the rectified signal to form a low frequency signal.

A low-frequency signal passes through a capacitor to form a demodulated signal. Fast Fourier transform is performed for the demodulated signal 116 to form a vibration spectrum. υδ 6053047 also involves calculating the resonance frequency of each physical path from the accelerometer to various sources of vibration in the electric motor, and υδ 6053047 involves performing this calibration step before the electric motor leaves the factory. Alternatively, such a calibration of each physical path from various vibration sources to the accelerometer should be performed using a calibrated hammer, according to υδ 6053047.

Summary of the invention

An aspect of the invention relates to a device for analyzing the condition of a machine having a part rotating at a speed of rotation, comprising a first sensor configured to generate an analog electrical measurement signal (δ_ΕΑ) depending on the mechanical vibrations resulting from the rotation of the said part;

analog-to-digital converter (44) for sampling an analog measuring signal at a frequency (Τ_δ) sampling so as to generate a signal (δ ^) of the digital measurement data in response to the received analog measurement data; moreover, the signal (δ ^) of the digital measurement data has a first signal-to-noise ratio;

first decimator for performing decimation of the signal (δ ^, δ_ΕΝν) digital measurement data so as to reach the first digital signal (δΜο, δΕΝ_ν) having a first reduced frequency (ί_δΕ1) discretization;

a second decimator (470, 470A, 470V), the second decimator (470, 470A, 470V) having a first input for receiving a first digital signal (BMO, δ_ΕΝν) and a second input for receiving a signal indicating a variable speed (ί_Ε0Τ) the rotation associated with the said part; a third input for receiving a signal indicating a signal for setting an output sample rate;

moreover, the second decimator (470, 470A, 470B) is configured to generate a second digital signal (δ_ΕΕϋ2) having a second reduced frequency (T_G2) discretization, in response to the first 021908 first digital signal (8ml, 8_ΕΝν), the signal indicates the corresponding speed (ί ^ οτ) of rotation, and the signal indicates the signal for setting the output sampling frequency so that the number of sample values ​​per revolution of the rotating part is kept equal to a practically constant value; and a signal-to-noise ratio enhancement module having an input for receiving a second digital signal (8_KEP2); moreover, the module for increasing the signal-to-noise ratio is configured to receive the first set (A ^ etn) of sample values, while the second digital signal (8_ΕΕϋ2) represents mechanical vibrations occurring as a result of rotation of the said part for a certain time;

moreover, the module for increasing the signal-to-noise ratio is configured to correlate so as to form a sequence (O) of output signals in which the component amplitudes of the repeating signals are amplified relative to the components of stochastic signals;

an evaluation module (230) for performing the state analysis function (P1, P2, Pu) for analyzing the state of the machine depending on the second digital signal (8_ΕΕϋ2).

According to an embodiment of the device, the first decimator is configured to reduce the sampling rate by an integer factor (M).

Aspect B1 of the invention relates to a computer program for instructing a computer to analyze a condition of a machine having a slowly rotating part, the computer program comprising computer-readable code means that, when executed on a computer, instructs the computer to generate an analog electrical measurement signal (8_EA) depending on the mechanical vibrations resulting from the rotation of the shaft;

means of a machine-readable code that, when executed on a computer, instructs the computer to sample the analog measuring signal at a frequency (ί₈) discretization so as to generate a signal (8mi) of digital measurement data in response to the received analog measurement data;

means of machine-readable code, which, when executed on a computer, instructs the computer to thin out the signal (8 ml) of digital measurement data so as to achieve a digital signal (8_ΕΕϋ) having a reduced frequency (G_G1, £_G2) discretization;

a computer-readable code tool that, when executed on a computer, instructs the computer to control the reduced frequency (£_8y, ί₈ι <₂) discretization so that the number of sample values ​​per shaft revolution (8) remains equal to an almost constant value; and a machine-readable code tool that, when executed on a computer, instructs the computer to perform a state analysis function (P1, P2, Pn) to analyze the state of the machine depending on the digital signal (8_ΚΕπ) having a reduced frequency (ί₈κι, ί₈ι <₂) discretization.

A computer program product comprising a computer-readable medium and a computer program according to aspect B1 of the claims, wherein the computer program is recorded on a computer-readable medium.

The invention also relates to a state monitoring system comprising a shock pulse measurement sensor associated with a planetary gear (700) for generating an analog signal;

an analog-to-digital converter connected to receive an analog signal; many functions (94, 240, 250, 310, 470, 320) of signal processing.

The invention also relates to a method for operating a filter with a finite impulse response having an input (480) for receiving detected input values ​​(8 (])) of signal data (8 ml) of digital measurement data depending on mechanical vibrations resulting from shaft rotation, wherein the signal (8ml) of digital measurement data has a frequency (£_8K1) discretization; and an input for receiving a signal indicating the rotational speed of the monitored rotating part at the time of the associated detection of input data values ​​(8 (])); and a storage device (604), configured to receive and store data values ​​(8 (])) and information indicating the corresponding speed (£_κοτ) rotation; and a value generator (606) configured to generate a fractional value (Ό); and a plurality of P1K filter taps having separate filter values; wherein the method comprises the step of interpolating the filter value.

The technical result to which the present invention is directed is to increase the reliability of detecting the state of a machine having a rotating part.

- 2 021908

Brief Description of the Drawings

The invention is further explained in the description of the preferred embodiments with reference to the accompanying drawings, in which the following is shown.

FIG. 1 is a schematic block diagram of an embodiment of a state analysis system 2 according to an embodiment of the invention.

FIG. 2A is a schematic block diagram of an embodiment of a part of the state analysis system 2 shown in FIG. one.

FIG. 2B is a schematic block diagram of an embodiment of an interface for sensors.

FIG. 2C shows a measurement signal from a vibration sensor.

FIG. 2Ό shows the amplitude of a measurement signal generated by a shock pulse sensor.

FIG. 2E depicts the amplitude of a measurement signal generated by a vibration sensor.

FIG. 3 is a simplified illustration of a shock pulse measurement sensor according to an embodiment of the invention.

FIG. 4 is a simplified illustration of an embodiment of a storage device 60 and its contents.

FIG. 5 is a schematic block diagram of an embodiment of an analysis device at a client location with a machine 6 having a movable shaft.

FIG. 6 is a schematic block diagram of an embodiment of a preprocessor according to an embodiment of the present invention.

FIG. 7 shows an embodiment of an evaluation module 230.

FIG. 8 depicts another embodiment of an evaluation module 230.

FIG. 9 depicts another embodiment of preprocessor 200.

FIG. 10A is a flowchart that illustrates embodiments of a method for improving repetitive signal patterns in signals.

FIG. 10B is a flowchart illustrating a method for generating a digital output signal.

FIG. 11 is a schematic illustration of a first storage device having several positions in a storage device.

FIG. 12 is a schematic illustration of a second storage device having several positions 1 in a storage device.

FIG. 13 is a schematic illustration of an example output signal 8ΜΌΡ containing two signatures of repeating signals.

FIG. 14A shows the number of sampled values ​​in a signal delivered to a decimator input

310.

FIG. 14B shows output sample values ​​of a corresponding time period.

FIG. 15A shows a decimator according to an embodiment of the invention.

FIG. 15B depicts another embodiment of the invention.

FIG. 16 depicts an embodiment of the invention including a decimator and a signal-to-noise ratio enhancement module, as described above, and a fractional decimator.

FIG. 17 shows an embodiment of a fractional decimator.

FIG. 18 depicts another embodiment of a fractional decimator.

FIG. 19 shows a decimator and another embodiment of a fractional decimator.

FIG. 20 is a block diagram of a decimator and another other embodiment of a fractional decimator.

FIG. 21 is a flowchart illustrating an embodiment of a method of working with a decimator and a decimator of decimation in FIG. twenty.

FIG. 22A, 22B, and 22C depict a method that can be implemented as a computer program.

FIG. 23 is a front view illustrating a planetary gear train.

FIG. 24 is a schematic side view of the planetary gear 700 of FIG. 23 when viewed in the direction of arrow 8 \ 7 in FIG. 23.

FIG. 25 depicts an analog version of an example signal generated by and output by preprocessor 200 (see FIGS. 5 or 16) in response to signals detected by at least one sensor 10 when the planetary gear rotates.

FIG. 26 shows an example of a portion of the high amplitude region 702A of the signal shown in FIG. 25.

FIG. 27 depicts an exemplary frequency spectrum of a signal containing a small periodic disturbance 903, as illustrated in FIG. 26.

FIG. 28 depicts an example of a portion of the example signal shown in FIG. 25.

- 3 021908

FIG. 29 depicts yet another embodiment of a state analysis system according to an embodiment of the invention.

FIG. 30 is a block diagram illustrating parts of the signal processing arrangement of FIG. 29 along with the user interface and display.

Detailed Description of Embodiments

In the following description, similar features in various embodiments may be indicated by identical reference numerals.

FIG. 1 shows a schematic block diagram of an embodiment of a state analysis system 2 according to an embodiment of the invention. Reference number 4 refers to a client location with a machine 6 having a moving part 8. The moving part may comprise bearings 7 and a shaft 8 that rotate when the machine is operating.

The operating state of the shaft 8 or the bearing 7 can be detected in response to vibrations emanating from the shaft and / or the bearing when the shaft rotates. Client location 4, which may also be referred to as the client part or user part, for example, may be the territory of a wind farm, i.e. a group of air turbines at a location or on the territory of a pulp and paper mill or some other manufacturing plant having machines with moving parts.

An embodiment of the state analysis system 2 is functional when the sensor 10 is attached at or at a measurement point 12 on the machine body 6. Although FIG. 1 illustrates only two measurement points 12, it should be understood that location 4 may contain any number of measurement points 12. The state analysis system 2 shown in FIG. 1, comprises an analysis device 14 for analyzing a machine condition based on measurement values ​​delivered by a sensor 10.

The analysis device 14 has a communication port 16 for bi-directional data exchange. The communication port 16 is connected to the communication network 18, for example, via the data interface 19. The communications network 18 may be a worldwide Internet network, also known as the Internet. The communication network 18 may also comprise a public switched telephone network.

The server computer 20 is connected to the communication network 18. Server 20 may comprise a database 22, user input / output interfaces 24 and data processing hardware 26 and a communication port 29. Server computer 20 is located at location 28, which is geographically separate from client location 4. Server location 28 can be located in the first city, for example, in the capital of Sweden, Stockholm, and the client location can be located in another city, for example, in Stuttgart, Germany, or in Detroit in Michigan, USA. Alternatively, the server location 28 may be located in the first part of the city, and the client location may be located in another part of the city. Server location 28 may also be referred to as provider part 28 or provider part location 28.

According to an embodiment of the invention, the central control room 31 comprises a control computer 33 having hardware and data processing software for monitoring a plurality of machines at a client location 4. The machines 6 may be air turbines or gearboxes used in air turbines. Alternatively, machines may include machinery, for example, in a pulp and paper mill. The control computer 33 may comprise a data base 22B, user input / output interfaces 24B and data processing hardware 26B and a communication port 29B. The central control room 31 may be separated from the client location 4 by a geographical distance. Via the communication port 29B, the control computer 33 may be connected to exchange data with the analysis device 14 through the port 16. The analysis device 14 may deliver measurement data partially processed so as to enable additional signal processing and / or analysis to be performed at a central location 31 via the control computer 33.

The supplier company occupies a location of 28 parts of the supplier. A supplier company may sell and deliver analysis devices 14 and / or software for use in analysis device 14. A supplier company may also sell and deliver analysis software for use in a control computer at a central control room 31. Such analysis software 94, 105 is explained in connection with FIG. 4 below. Such analysis software 94, 105 may be delivered via transmission over said communication network 18.

According to one embodiment of system 2, device 14 is a portable device that can periodically connect to communication network 18.

According to another embodiment of the system 2, the device 14 is connected to the communication network 18 essentially continuously. Therefore, the device 14 according to this embodiment may almost always be available online for communication with the supplier computer 20 and / or with the control computer 33 in the control room 31.

FIG. 2A is a schematic block diagram of an embodiment of a part of the state analysis system 2 shown in FIG. 1. A state analysis system, as illustrated in FIG. 2A, contains a sensor module 10 4 021908 for generating a measured value. The measured value may depend on displacement or, more precisely, may depend on vibrations or shock pulses caused by bearings as the shaft rotates.

An embodiment of the state analysis system 2 is functional when the device 30 is rigidly mounted on or at the measurement point on the machine 6. The device 30 mounted on the measurement point may be referred to as a pin 30. The pin 30 may comprise a connector 32 to which the sensor module 10 joins in a removable manner. The coupler 32, for example, may include a dual-thread to allow mechanical engagement of the sensor module with the journal by 1/4 turn.

The measurement point 12 may comprise a threaded groove in the machine body. The trunnion 30 may have a protruding part with a thread corresponding to the grooves, to enable the trunnion to be firmly attached to the measuring point by inserting, for example, a bolt into the groove.

Alternatively, the measurement point may comprise a threaded groove in the machine body, and the sensor module 10 may comprise a corresponding thread so that it can be inserted directly into the groove. Alternatively, the measuring point is marked on the machine body only with a colored mark.

The machine 6 illustrated in FIG. 2A may have a rotating shaft with a defined shaft diameter Y1. The shaft in machine 24 can rotate at a speed VI of rotation when machine 6 is in use.

The sensor module 10 may be connected to the machine status analysis device 14. With reference to FIG. 2A, the analysis device 14 comprises a sensor interface 40 for receiving a measured signal or measurement data generated by the sensor 10. The sensor interface 40 is connected to data processing means 50 capable of controlling the operation of the analysis device 14 in accordance with a program code. The data processor 50 is also connected to a storage device 60 for storing program code.

According to an embodiment of the invention, the sensor interface 40 includes an input 42 for receiving an analog signal, the input 42 being connected to an analog-to-digital (A / A) converter 44, the digital output of which 48 is connected to the data processing means 50. A / D converter 44 samples the received analog signal at a specific frequency ί₃ sampling so as to deliver a digital measurement data signal having a specific frequency ί₃ sampling, the amplitude of each sample depends on the amplitude of the received analog signal at the time of sampling.

According to another embodiment of the invention illustrated in FIG. 2B, the sensor interface 40 includes an input 42 for receiving an analog signal §_EA from a shock pulse measurement sensor, a deduction circuit 43 connected to receive an analog signal, and an analog-to-digital converter 44 connected to receive an analogue signal reduced to a demanded parameter from a deduction circuit 43. An analog-to-digital converter 44 samples the received analogue signal with a certain frequency ί₃ sampling so as to deliver a digital measurement data signal having a specific frequency ί₃ sampling, the amplitude of each sample depends on the amplitude of the received analog signal at the time of sampling.

The discrete representation theorem ensures that signals with a limited frequency band (i.e., signals having a maximum frequency) can ideally be restored from their discretized version if the sampling frequency более is more than twice the maximum frequency T_EAanalog signal tach 8_eato be tracked. A frequency equal to half the sampling frequency is therefore the theoretical limit of the highest frequency that can be unambiguously represented by a sampled signal §mo. This frequency (half the sampling frequency) is called the Nyquist frequency of the sampling system. Frequencies above frequency ί_Ν Nyquist can be observed in a sampled signal, but their frequency is ambiguous. Those. the frequency component with frequency ί cannot differ from other components with frequencies Βχί_Ν+ ί and Βχί_Ν-ί for non-zero integers B. This ambiguity, known as spectral overlay, can be processed by filtering the signal using a filter to smooth out the overlay (usually a low-pass filter with a cut-off near the Nyquist frequency) before converting to a discrete discrete representation.

In order to provide a margin of safety from the point of view of allowing an imperfect filter to have a certain slope in the frequency response, the sampling frequency can be chosen equal to a value greater than 2. Therefore, according to embodiments of the invention, the sampling frequency can be set equal to:

<img file="EA021908B1_D0001.tif" />

where k is a coefficient having a value greater than 2.0.

Accordingly, the coefficient k can be selected equal to a value in excess of 2.0. Preferably, the coefficient k can be chosen to be between 2.0 and 2.9 in order to provide a good safety margin while avoiding the formation of too many sample values. According to an embodiment, the coefficient k is advantageously selected such that 100xx / 2 provides an integer. According to an embodiment, the coefficient k may be set to 2.56. The choice of equal to 2.56 provides 100hk = 256 = 2 to the power of 8.

According to an embodiment, the frequency G₃ signal sampling_mi digital measurement data can be set in a fixed way equal to a certain value of GZ, such as, for example, G_s= 102 kHz.

Therefore, when the frequency₃ discretization is set in a fixed way equal to a certain value of₃maximum frequency G_Zeaht analog signal Z_EA is £$ ЕА out⁼Gg / k where G_Zeaht is the highest frequency that should be analyzed in a sampled signal.

Therefore, when the frequency of the GB sampling is set in a fixed way equal to a certain value of G_s= 102400 Hz, and the coefficient k is set equal to 2.56, the maximum frequency G_Zeahtanalog signal Z_EA makes up

G5EAshah⁼G £ / k = 102400 / 2.56 = 40 kHz

Accordingly, signal Z_{m |}.) digital measurement data having a specific frequency G_s sampling, is formed in response to the received analog measuring signal Z_EA. The digital output 48 of the analog-to-digital converter 44 is connected to the data processing means 50 through the output 49 of the sensor interface 40 so as to deliver signal Z_{m |}.) digital measurement data to data processing means 50.

The sensor module 10 may comprise a vibration sensor, the sensor module having a structure so as to physically engage the coupling of the measurement point so that the vibrations of the machine at the measurement point are transferred to the vibration sensor. According to an embodiment of the invention, the sensor module comprises a transducer having a piezoelectric element. When the measurement point 12 vibrates, the sensor module 10, or at least a portion thereof, also vibrates, and the converter then generates an electrical signal whose frequency and amplitude depend on the frequency of the mechanical vibration and the vibration amplitude of the measurement point 12, respectively. According to an embodiment of the invention, the sensor module 10 is a vibration sensor providing an analog amplitude signal, for example, 10 mV / g in the frequency range 1.00-10000 Hz. Such a vibration sensor is configured to deliver an almost identical amplitude of 10 mV, regardless of whether an acceleration of 1 d is applied to it (9.82 m / s²) at 1, 3 or 10 Hz. Therefore, a typical vibration sensor has a linear response in the indicated frequency range up to about 10 kHz. Mechanical vibrations in this frequency range emanating from the rotating parts of the machine are usually caused by imbalance or inaccurate alignment. However, when mounted on a machine, a linear vibration characteristic sensor typically also has several different mechanical resonant frequencies depending on the physical path between the sensor and the vibration source.

Damage in the roller bearing leads to relatively sharp elastic waves, known as shock pulses, passing along the physical path in the machine body until the sensor is reached. Such shock pulses often have a wide frequency spectrum. The amplitude of the shock pulse of the roller bearing is typically lower than the amplitude of the vibration caused by imbalance or inaccurate alignment.

The wide frequency spectrum of shock pulse signatures allows them to activate a loop response or resonance at the resonant frequency associated with the sensor. Therefore, a typical measurement signal from a vibration sensor may be in the form of a signal, as shown in FIG. 2C, i.e. dominant low-frequency signal with a superimposed resonant loop characteristic with a higher frequency and lower amplitude.

To provide an analysis of the signature of the shock pulses, often resulting from bearing damage, the low-frequency component must be filtered out. This can be achieved by a high-pass filter or by a band-pass filter. However, these filters must be adjusted so that part of the low-frequency signals is blocked, while part of the high-frequency signals passes. A single vibration sensor typically has one resonant frequency associated with the physical path from one source of the shock pulse signal and another resonant frequency associated with the physical path from the other source of the shock pulse signal, as mentioned in IZ 6053047. Therefore, filter control aimed at skipping parts of high-frequency signals, requires separate adaptation when a vibration sensor is used.

When such a filter is correctly configured, the resulting signal consists of a shock pulse signature. However, the analysis of the signature of the shock pulses emanating from the vibration sensor is violated to some extent by the fact that the amplitude response, as well as

- 6 021908 resonant frequency, in fact, vary depending on the individual physical path from the signal sources of the shock pulses.

Advantageously, these disadvantages associated with vibration sensors can be reduced by a shock pulse measurement sensor. The shock pulse measurement sensor is configured to provide a predetermined mechanical resonant frequency, as described in more detail below.

This feature of the shock pulse measurement sensor predominantly makes the measurement results repeatable in that the output signal from the shock pulse measurement sensor has a stable resonant frequency that is virtually independent of the physical path between the shock pulse signal source and the shock pulse sensor. In addition, mutually different individual shock pulse sensors provide a very small, if any, deviation in the resonant frequency.

The advantage of this is that the signal processing is simplified due to the fact that the filters do not have to be configured separately, in contrast to the case described above when vibration sensors are used. In addition, the amplitude response from the shock pulse sensors is clearly defined so that a single measurement provides reliable information when the measurement is performed in accordance with the appropriate measurement methods specified by δ.Ρ.Μ. 1 and 81gitep! AB.

FIG. 2Ό illustrates the amplitude of a measurement signal generated by a shock pulse sensor, and FIG. 2E illustrates the amplitude of a measurement signal generated by a vibration sensor. Both sensors are applied to an identical mechanical shock sequence without the typical contents of a low-frequency signal. As clearly seen in FIG. 2Ό and 2E, the duration of the resonance characteristic for the shock pulse signature from the shock pulse measurement sensor is less than the corresponding resonance characteristic for the shock pulse signature from the vibration sensor. This feature of a shock pulse measurement sensor in the form of providing various characteristics with shock pulse signatures has the advantage of providing a measurement signal from which various mechanical shock pulses that arise in a short period of time can be distinguished.

According to an embodiment of the invention, the sensor is a shock pulse measurement sensor. FIG. 3 is a simplified illustration of a shock pulse measurement sensor 10 according to an embodiment of the invention. According to this embodiment, the sensor comprises a portion 110 having a certain mass or weight and a piezoelectric element 120. The piezoelectric element 120 is somewhat flexible so that it can contract and expand when an external force is applied. The piezoelectric element 120 comprises electrically conductive layers 130 and 140, respectively, on opposite surfaces. As the piezoelectric element 120 contracts and expands, it generates an electrical signal that is picked up by the conductive layers 130 and 140. Accordingly, mechanical vibration is converted into an analog electrical measurement signal 8EA delivered to the output terminals 145, 150.

The piezoelectric element 120 can be placed between the weight 110 and the surface 160, which, during operation, is physically connected to the measurement point 12, as illustrated in FIG. 3.

The shock pulse measuring sensor 10 has a resonant frequency depending on the mechanical characteristics of the sensor, for example, the mass t of the weight part 110 and the elasticity of the piezoelectric element 120. Therefore, the piezoelectric element has elasticity and spring stiffness. Mechanical resonant frequency G_Km for the sensor, therefore, also depends on the mass t and the stiffness to the spring.

According to an embodiment of the invention, the mechanical resonant frequency G_Km for the sensor can be determined by the following equation:

<img file="EA021908B1_D0002.tif" />

According to another embodiment, the actual mechanical resonance frequency for the shock pulse measurement sensor 10 may also depend on other factors, for example, the nature of the attachment of the sensor 10 to the machine body 6.

The resonance sensor 10 for measuring shock pulses thereby, in particular, is sensitive to vibrations having a frequency at or near the mechanical resonant frequency G_Km. The sensor 10 for measuring shock pulses can be made so that the mechanical resonant frequency G_Kmis approximately in the range from 28 to 37 kHz. According to another embodiment, the mechanical resonant frequency G_Km is approximately in the range from 30 to 35 kHz.

Accordingly, the analog electrical measurement signal has an electrical amplitude that can vary over the frequency spectrum. For the purpose of describing the theoretical foundations, we can assume that if mechanical vibrations with the same amplitude at all frequencies, for example, from 1, for example, to 200,000 kHz, are applied to the sensor 10 for measuring shock pulses, then the amplitude of the analog signal 8_EA from the shock pulse measurement sensor should have a maximum at a mechanical resonant frequency Г_Km. since the sensor resonates when pushed at this frequency.

- 7 021908

With reference to FIG. 2B, the circuit 43 for bringing to the required parameters receives an analog signal §еа. The circuit 43 for bringing to the required parameters can be performed as an impedance adaptation circuit configured to adapt the input impedance of an analog-to-digital converter, as can be seen from the contact terminals of the sensor 145, 150, so that optimal signal transmission is carried out. Therefore, the circuit 43 to bring to the desired parameters can be performed with the ability to adapt the input impedance Ζ_ιηas can be seen from the contact terminals of the sensor 145, 150, so that the maximum electric power is delivered to the analog-to-digital converter 44. According to an embodiment of the circuit 43 for bringing to the required parameters, the analog signal 8_ea fed into the primary winding of the transformer, and the analog signal reduced to the required parameters is delivered by means of the secondary winding of the transformer. The primary winding has η1 turns, and the secondary winding has η2 turns, the ratio η1 / η2 = η12. Therefore, the analog-to-digital converter 44 is connected to receive the analog signal brought to the required parameters from the circuit 43 for bringing to the required parameters. Analog-to-digital converter 44 has an input impedance Ζ₄₄, and the input impedance of the analog-to-digital converter, as can be seen from the contact terminals of the sensor 145, 150, is (η1 / η2)²χΖ₄₄when the bringing circuit 43 to the required parameters is connected between the contact terminals of the sensor 145, 150 and the input terminals of the analog-digital converter 44.

An analog-to-digital converter 44 samples the received analogue signal with a certain frequency ί₃ sampling so as to deliver a digital measurement data signal having a specific frequency ί₃ sampling, the amplitude of each sample depends on the amplitude of the received analog signal at the time of sampling.

According to embodiments of the invention, the signal §mo of the digital measurement data is delivered to the digital signal processing means 180 (see FIG. 5).

According to an embodiment of the invention, the digital signal processing means 180 comprises a data processor 50 and program code for instructing the data processor 50 to process the digital signals. According to an embodiment of the invention, the processor 50 is implemented by a digital signal processor. A digital signal processor may also be referred to as ΌδΡ.

With reference to FIG. 2A, the data processing means 50 is connected to a storage device 60 for storing program code. The program memory 60 is preferably non-volatile memory. The storage device 60 may be random access memory, i.e. providing the ability to both read data from the storage device and write new data to the storage device 60. According to an embodiment, the program memory 60 is implemented via flash memory. The program storage device 60 may include a first storage device segment 70 for storing a first set of program code 80 that is configured to control the analysis device 14 to perform basic operations (FIGS. 2A and 4). The program memory may also include a second memory segment 90 for storing a second set of program code 94. The second set of program code 94 in the second memory segment 90 may include program code for instructing the analysis device to process a specific signal or signals so as to generate a pre-processed signal or a set of pre-processed signals. The storage device 60 may also include a third storage segment 100 for storing a third set of program code 104. The set of program code 104 in the third storage segment 100 may include program code for instructing the analysis device to perform a selected analysis function 105. When the analysis function is performed, it can instruct the analysis device to present the corresponding analysis result in the user interface 106 or deliver the analysis result to port 16 (see FIGS. 1 and 2A and 7 and 8).

The data processing means 50 is also connected to the random access memory 52 for storing data. In addition, the data processing means 50 may be connected to the communication interface 54 of the analysis device. The communication interface 54 of the analysis device provides bi-directional communication with the communication interface 56 of the measurement point, which is connected to, at or near the measurement point on the machine.

The measurement point 12 may include a connector 32, a read and write storage medium 58, and a measurement point communication interface 56.

A recordable information medium 58 and a measurement point communication interface 56 may be provided in a separate device 59 located near the journal 30, as illustrated in FIG. 2. Alternatively, a recordable storage medium 58 and a communication interface 56 for the measurement point can be provided in the journal 30. This is described in more detail in publication \ UO 98/01831, the contents of which are incorporated herein by reference.

System 2 is configured to provide bi-directional communication between the measurement point communication interface 56 and the analysis device communication interface 54. Measurement Point Communication Interface 56

- 8 021908 and the communication interface 54 of the analysis device is preferably configured to provide wireless communication. According to an embodiment, the measurement point communication interface and the analysis device communication interface are configured to exchange data with each other by means of radio frequency (KR) signals. This embodiment includes an antenna in the measurement point communication interface 56 and another antenna in the analysis device communication interface 54.

FIG. 4 is a simplified illustration of an embodiment of a storage device 60 and its contents. The simplified illustration intends to convey an understanding of the general idea of ​​storing various program functions in the storage device 60, and it is not necessarily the correct technical idea of ​​the way the program should be stored in a real storage device circuit. The first storage segment 70 stores program code for controlling the analysis device 14 to perform basic operations. Although the simplified illustration of FIG. 4 shows pseudo code, it should be understood that program code 80 may consist of machine code or program code of any level that can be executed or interpreted by means of processing means 50 (FIG. 2A).

The second storage segment 90, illustrated in FIG. 4 stores the second set of program code 94. The program code 94 in segment 90, when executed in the data processor 50, instructs the analysis device 14 to perform a function, for example, a digital signal processing function. The function may include improved mathematical processing of the signal δ_Μυdigital measurement data. According to embodiments of the invention, program code 94 is configured to instruct processor means 50 to perform signal processing functions described in connection with FIG. 5, 6, 9 and / or 16 in this document.

As mentioned above in connection with FIG. 1, a computer program for controlling the function of the analysis device may be downloaded from the server computer 20. This means that the program to be downloaded is transmitted over the communication network 18. This can be done by modulating the carrier in order to carry the program over the communication network 18. Accordingly, the downloaded program can be downloaded to a digital storage device, for example, a storage device 60 (see FIGS. 2A and 4). Therefore, the signal processing program 94 and or the analytic functions program 104, 105 may be received through a communication port, for example, port 16 (FIGS. 1 and 2A) so as to load it into the storage device 60. Similarly, the signal processing program 94 and / or the program of analytical functions 104, 105 can be received through the communication port 29B (Fig. 1) so as to download it to the location of the program storage device in the computer 26B or in the database 22B.

An aspect of the invention relates to a computer program product, for example, program code means 94 and / or program code means 104, 105, which are downloaded to a digital storage device. A computer program product containing portions of program code for performing signal processing methods and / or analytical functions when the product is executed in a data processing unit 50 of a device for analyzing a machine state. The term is executed in the data processing module means that the computer program plus the data processing module performs a method of the type described in this document.

The wording of a computer program product loaded into a digital memory of a state analysis device means that a computer program can be entered into a digital memory of a state analysis device so as to obtain a state analysis device programmed, adapted, or configured to implement a method of the type described above. The term loaded into the digital storage device of the state analysis device means that the state analysis device programmed in this way allows or is configured to implement a method of the type described above.

The above computer program product may also be downloaded to a computer-readable medium, for example, a CD or an EEE. Such a machine-readable medium may be used to deliver a program to a client.

According to an embodiment of the analysis device 14 (FIG. 2A), it comprises a user input interface 102 through which an operator can interact with the analysis device 14. According to an embodiment, the user input interface 102 comprises a set of buttons 104. An embodiment of the analysis device 14 comprises a user output interface 106. The user output interface may include a display 106. The data processor 50, when it performs the basic program function provided in the base program code 80, provides user interaction through the user input interface 102 and the display 106. The set of buttons 104 may be limited to a few buttons, for example, five buttons, as illustrated in FIG. 2A. The central button 107 can be used for the ΕΝΤΕΚ or §ЕЕЕСТ function, while other additional peripheral buttons can be used to move the cursor on the display 106. Thus, it should be understood that characters and text can be entered into the device 14 via the user interface. The display 106, for example, can display a certain number of characters, for example, letters of the alphabet, while the cursor moves on the display in response to user input so as to allow the user to enter information.

- 9 021908

FIG. 5 is a schematic block diagram of an embodiment of an analysis device 14 at a client location 4 with a machine 6 having a movable shaft 8. A sensor 10, which may be a shock pulse measurement sensor, is shown attached to the body of the machine 6 so as to remove mechanical vibrations, and so on to deliver the analog measurement signal δ_ΕΑindicating certain mechanical vibrations to interface 40 for sensors. The sensor interface 40 may be configured as described in connection with FIG. 2A or 2B. The interface 40 for the sensors delivers the signal δ_Μϋ digital measurement data to means 180 for processing digital signals.

Signal δ_Μϋ digital measurement data has a frequency Γ_δ sampling, and the amplitude value of each sample depends on the amplitude of the received analog measuring signal δ_ΕΑ at the time of sampling. According to an embodiment, the frequency Γ_δ signal sampling δ_Μϋ digital measurement data can be set in a fixed way equal to a certain value of ίδ, such as, for example, ί_δ= 102 kHz. Frequency Γ_δ sampling may be controlled by a clock delivered by a clock 190, as illustrated in FIG. 5. The clock signal may also be delivered to means 180 for processing digital signals. Means 180 for processing digital signals can generate information about the time duration of the received signal δ_Μϋ digital measurement data in response to the received signal δ_Μϋ digital measurement data, the clock signal and the relationship between the frequency Γ_δ sampling and clock since the duration between two consecutive sampled values ​​is Τ_δ= 1 / Γ_δ.

According to embodiments of the invention, the means 180 for processing digital signals includes a preprocessor 200 for performing preprocessing of the signal δ_Μϋ digital measurement data so as to deliver a pre-processed digital signal δ_ΜϋΡ output 210. Output 210 is connected to input 220 of evaluation module 230. The evaluation module 230 is configured to evaluate a pre-processed digital signal δ_ΜϋΡ so as to deliver the evaluation result to the user interface 106. Alternatively, the evaluation result may be delivered to the communication port 16 so as to provide transmission of the result, for example, to the control computer 33 on the control node 31 (see Fig. 1).

According to an embodiment of the invention, the functions described in connection with the functional blocks in the digital signal processing means 180, the preprocessor 200 and the evaluation unit 230, can be implemented by computer program code 94 and / or 104, as described in connection with the storage units 90 and 100 devices in connection with the above FIG. four.

The user may require only a few basic monitoring functions to determine if the condition of the machine is normal or abnormal. When determining the abnormal condition, the user can call professional service and repair specialists to establish the exact nature of the problem, as well as to perform the necessary maintenance and repair.

Professional maintenance and repair professionals often must and use a wide range of evaluation functions to establish the nature and / or cause of the abnormal condition of the machine. Therefore, various users of the analysis device 14 can lead to significantly different needs for device functions. The term condition monitoring function is used in this document for a function to determine whether the condition of the machine is normal or to a certain extent deteriorated or abnormal. The term condition monitoring function also includes an evaluation function that allows one to establish the nature and / or cause of the abnormal condition of the machine.

Examples of machine status monitoring functions.

The functions P1, P2, ..., Pu of state monitoring include such functions as vibration analysis, temperature analysis, shock pulse measurement, spectral analysis of shock pulse measurement data, fast Fourier transform of vibration measurement data, graphical representation of state data in the user interface storing state data on a recordable information medium on a machine, storing state data on a recordable information medium in a device, tachometry, detecting imbalance and detecting inaccurate alignment.

According to an embodiment, the device 14 includes the following functions: P1 = vibration analysis; P2 = temperature analysis; P3 = shock pulse measurement; P4 = spectral analysis of shock pulse measurement data; P5 = fast Fourier transform of vibration measurement data; P6 = graphical representation of status data in the user interface; P7 = storage of status data on a recordable storage medium on a machine; P8 = storing status data on a recordable information medium 52 in a device; P9 = tachometry; P10 = detect imbalance and P11 = detect inaccurate alignment; P12 = extracting status data from a recordable information medium 58 on a machine; P13 = performing function P1 of vibration analysis and performing function P12 Extract status data from a recordable information medium 58 on a machine; in order to provide a comparison or trend analysis based on current vibration measurement data and vibration measurement statistics; P14 = Perform P2 temperature analysis; and performing a function of extracting state data from a recordable information medium 58 on a machine; so as to

- 10 021908 to provide a comparison or analysis of trends based on current temperature measurement data and temperature measurement statistics; P15 = retrieval of identification data from the recordable information medium 58 on the machine.

Embodiments of function P7 storing state data on a recordable information medium on a machine and P13 vibration analysis and extracting state data are described in more detail in \ νϋ 98/01831, the contents of which are contained in this document by reference.

FIG. 6 illustrates a schematic block diagram of an embodiment of a preprocessor 200 according to an embodiment of the present invention. In this embodiment, the signal §_msdigital measurement data is connected to a digital band-pass filter 240 having a lower frequency_{The EU} cut-off, upper frequency G_is cutoff and bandwidth between the upper and lower cutoff frequencies.

The output of the digital band-pass filter 240 is connected to a digital envelope shaping module 250. According to an embodiment of the invention, the signal output from the envelope forming unit 250 is delivered to the output 260. The output 260 of the preprocessor 200 is connected to the output 210 of the digital signal processing means 180 for delivery to the input 220 of the evaluation unit 230.

The upper and lower cutoff frequencies of the digital band-pass filter 240 can be selected so that the frequency components of the signal δ_Μυ at the resonant frequency for the sensor are in the bandwidth. As mentioned above, the amplification of mechanical vibration is achieved by a sensor mechanically resonating at the resonant frequency G_{| m}. Accordingly, the analog measurement signal 8ea reflects the amplified value of the vibrations at and near the resonant frequency. Therefore, the bandpass filter according to the embodiment of FIG. 6 predominantly suppresses the signal at frequencies lower and higher than the resonant frequency ksh, so as to further improve the components of the measuring signal at the resonant frequency Гш- In addition, the digital bandpass filter 240 mainly further reduces noise, which is essentially included in the measuring signal, since any noise components are lower than frequency g_bc cutoff and above the upper frequency g_iscut-offs are also eliminated or reduced. Therefore, when using the resonant sensor 10 for measuring shock pulses having a mechanical resonant frequency in the range from the lowest value of the resonant frequency to the highest value of_{| &lt;m} resonant frequency, the digital band-pass filter 240 can be designed to have a lower cutoff frequency_FROM= Gkm and the upper cutoff frequency G_is= R_{| mi}. According to an embodiment, the lower cutoff frequency Gb_FROM= Gcm = 28 kHz, and the upper cutoff frequency G_is= R_{| mi}= 37 kHz.

According to another embodiment, the mechanical resonant frequency G_Km is in the range of about 30 to 35 kHz, and the digital band-pass filter 240 can then be designed to have a lower cutoff frequency G_bc= 30 kHz and upper cutoff frequency G_is= 35 kHz.

According to another embodiment, the digital band-pass filter 240 may be designed to have a lower frequency G_bc cut-off below the lowest value of G_{| m} resonant frequency, and the upper frequency G_is cut-off exceeding the highest value of G_{| mi} resonant frequency. For example, the mechanical resonant frequency G_{| m} can be a frequency of 30 to 35 kHz, and the digital band-pass filter 240 in this case can be designed to have a lower cutoff frequency G_bc= 17 kHz and upper cutoff frequency G_is= 36 kHz.

Accordingly, the digital band-pass filter 240 delivers the band-pass signal 8_R digital measurement data having a predominantly low noise content and reflecting mechanical vibrations in the passband. Band signal 8_R digital measurement data is delivered to envelope shaping unit 250.

The digital envelope shaping module 250 accordingly receives the bandpass signal 8_Rdigital measurement data, which can reflect a signal having positive as well as negative amplitudes. With reference to FIG. 6, the received signal is rectified by a digital rectifier 270, and the rectified signal can be filtered by an optional low-pass filter 280 so as to form a digital signal δ_ΕΝν envelope.

Accordingly, the signal δ_ΕΝν is a digital representation of the envelope signal generated in response to the filtered signal 8_R measurement data. In some embodiments, an optional lowpass filter 280 may be omitted. One such embodiment is explained in connection with FIG. 9 below. Accordingly, an optional low-pass filter 280 in the envelope shaping unit 250 may be omitted when the decimator 310 explained in connection with FIG. 9 below includes a low pass filter function.

According to the embodiment of FIG. 6 inventions signal δ_ΕΝν delivered to the output 260 of preprocessor 200. Therefore, according to an embodiment of the invention, the pre-processed digital signal §_msrdelivered to output 210 (FIG. 5) is a digital signal δ_ΕΝν envelope.

While analog devices of the prior art use an analog rectifier to generate an envelope signal in response to a measurement signal, which, in essence, leads to the input of a bias error into the resulting signal, the digital envelope generation module 250 mainly generates a real rectification without bias errors . Accordingly, the digital signal δ_ΕΝν the envelope must have a good signal-to-noise ratio, since a sensor mechanically resonating at a resonant frequency in the passband of the digital band-pass filter 240 leads to a high amplitude of the signal, and signal processing performed in the digital domain eliminates the addition of noise and eliminates the addition of bias errors.

With reference to FIG. 5 pre-processed digital signal δ_ΜυΡ delivered to input 220 of evaluation module 230.

According to another embodiment, the filter 240 is a high-pass filter having a frequency of_bc cutoffs. This embodiment simplifies the design by replacing the band-pass filter with a high-pass filter 240, thereby leaving low-pass filtering for another downstream low-pass filter, for example, a low-pass filter 280. Frequency g_bc the cut-off of the high-pass filter 240 is selected approximately equal to the value of the lowest expected value T | &lt;_M1. · Mechanical resonant frequency of the resonant sensor 10 measuring shock pulses. When the mechanical resonant frequency Gsh is in the range of about 30 to 35 kHz, the high-pass filter 240 can be designed to have a lower cutoff frequency G_bc= 30 kHz. The high-pass filtered signal is then transmitted to a rectifier 270 and to a low-pass filter 280. According to an embodiment, it should be possible to use sensors 10 having a resonant frequency in the range of about 20 to 35 kHz. To achieve this, the high-pass filter 240 may be designed to have a lower cutoff frequency G_Bc = 20 kHz.

FIG. 7 illustrates an embodiment of an evaluation module 230 (see also FIG. 5). The embodiment of FIG. 7, evaluation module 230 includes a state analyzer 290 configured to receive a pre-processed digital signal δ_Μυ_Ρ· Indicating the state of the machine 6. The state analyzer 290 may be controlled to perform the selected state analysis function by means of a selection signal delivered to the control signal input 300. A selection signal delivered to control signal input 300 may be generated by user interaction with user interface 102 (see FIG. 2A). When the selected analytic function includes a fast Fourier transform, the analyzer 290 is defined by the selection signal 300 so that it controls the input signal in the frequency domain.

Depending on which analysis is to be performed, the state analyzer 290 may control the input pre-processed digital signal δ_ΜυΡ in the time domain or by a pre-processed digital signal δ_Μυ_Ρ in the frequency domain. Accordingly, depending on the selection signal delivered to the control signal input 300, the PPT 294 may be turned on as shown in FIG. 8, or signal δ_Ρ|_{Ρ) Ρ} can be delivered directly to analyzer 290, as illustrated in FIG. 7.

FIG. 8 illustrates another embodiment of an evaluation module 230. In the embodiment of FIG. 8, evaluation module 230 includes an optional fast Fourier transform module 294 connected to receive a signal from input 220 of evaluation module 230. The output from the PPT conversion module 294 may be delivered to the analyzer 290.

In order to analyze the state of a rotating part, it is necessary to track certain vibrations for a sufficiently long time to be able to detect repetitive signals. Certain signatures of repeating signals are indicative of a deteriorated state of the rotating part. Signature analysis of repetitive signals can also be a sign of a type of deteriorated condition. Such an analysis can also lead to the determination of the degree of deterioration.

Therefore, the measuring signal may include at least one component of the vibration signal depending on the vibrational movement of the rotationally moving part 8; the component of the vibration signal has a frequency of_from repetition depending on speed G_K0T rotation of the rotationally moving part 8. The component of the vibration signal, which depends on the vibrational movement of the rotationally moving part 8, can therefore serve as a sign of deteriorated condition or damage to the machine being tracked. In fact, the relationship between the frequency G_R) repetition of component δ_υ vibration signal and speed G_K0T rotation of the rotationally moving part 8 can serve as a sign of what kind of mechanical part is damaged. Therefore, in a machine having a plurality of rotating parts, it may be possible to identify a single slightly damaged part by processing the measurement signal using an analytic function 105 including frequency analysis.

Such a frequency analysis may include a fast Fourier transform of a measurement signal including component δ_Β vibration signal. Fast Fourier Transform (PPT) uses a specific frequency resolution.

A certain frequency resolution, which can be expressed in units of frequency resolution elements, defines a limit for distinguishing between different frequencies. The term frequency resolution elements is sometimes referred to as strings. If a frequency resolution is required that provides 0 elements of frequency resolution up to the shaft rotation speed, it is necessary to record the signal during X shaft revolutions.

In connection with the analysis of parts of rotation, it may be interesting to analyze signal frequencies that exceed the frequency £_Cat rotation of the rotating part. The rotating part may include a shaft and bearings. Frequency £_Cat shaft rotation is often referred to as order 1. Bearing signals of interest can occur about ten times per shaft revolution (order 10), i.e. damage repetition rate (measured in Hz) divided by the speed £_Cat rotation (measured in r / s) is 10 Hz / r / s, i.e. order y = r_about/ G_to._From= 10 Hz / r / s. In addition, it may be interesting to analyze the harmonics of the bearing signals, so it may be interesting to measure up to about 100. Referring to the maximum order as Υ and the total number of frequency resolution elements in PPT to be used as Ζ, the following applies: Ζ = ΧχΥ. On the contrary, Χ = Ζ / Υ, where

X is the number of revolutions of the tracked shaft during which the digital signal is analyzed;

Υ is the maximum order and

Ζ is the frequency resolution, expressed as the number of frequency resolution elements.

Consider the case where the thinned digital measurement signal 8 \ | [l- (see FIG. 5) is delivered to the RFC analyzer 294, as described in FIG. 8. In this case, when the PPT analyzer 294 is set for Ζ = 1600 frequency resolution elements and the user is interested in frequency analysis up to the order of Υ = 100, then the value for X becomes Χ = Ζ / Υ = 1600/100 = 16.

Therefore, it is necessary to measure during X = 16 shaft revolutions, when Ζ = 1600 frequency resolution elements are required, and the user is interested in frequency analysis up to the order of Υ = 100.

The frequency resolution Ζ of the PPT analyzer 294 can be set using the user interface 102, 106 (FIG. 2A).

Therefore, the frequency resolution value Ζ for the state analysis function 105 and / or the signal processing function 94 (FIG. 4) can be set using the user interface 102, 106 (FIG. 2A).

According to an embodiment of the invention, the frequency resolution Ζ is set by selecting one value Ζ from the group of values. The group of selected values ​​for frequency resolution Ζ may include

Ζ = 400,

Ζ = 800,

Ζ = 1600,

Ζ = 3200,

Ζ = 6400.

As mentioned above, the frequency £_δ discretization can be set in a fixed way equal to a certain value, such as, for example, £_δ= 102400 kHz, and the coefficient k can be set equal to 2.56, thereby setting the maximum frequency that should be analyzed, Γ_δ|._ΛιΙΙ,_ι;,. as ^ 8ELtax = £ 's / k = 102400 / 2.56 = 40 kHz.

For a machine having a shaft with a rotation speed T_Cat= 1715 rpm = 28.58 rpm, the selected value of the order Υ = 100 sets the maximum frequency, which should be analyzed as_KO^ = 28.58 rpm 100 = 2858 Hz.

The PPT transform module 294 may be configured to perform fast Fourier transform for a received input signal having a certain number of sample values. An advantageous situation is when a certain number of sample values ​​is set equal to an even integer, which can be divided into two (2) without providing a fractional number.

Accordingly, a data signal representing mechanical vibrations resulting from shaft rotation may include patterns of repeating signals. A certain signal pattern can thereby be repeated a certain number of times per revolution of the shaft being tracked. In addition, repetitive signals can occur with mutually different repetition rates.

In the work of Masyetu UfgaOop Meazitiez apb Aia1u81z of the author U1c1og \ Uo \\ k (ΙδΒΝ 0-07-071936-5), several examples of mutually different repetition frequencies are given on p. 149: main group frequency (RGR), ball rotation frequency (Βδ), outer ring (OK), inner ring (ΙΚ).

The work also provides formulas for calculating these specific frequencies on p. 150. The contents of the work of Masuietu Ufabot Meazitetzz apb Ai1u81z of the author Uyuyug \ Yo \\ k are contained in this document by reference. In particular, the above formulas for calculating these specific frequencies are contained herein by reference. Table on p. 151 of this work shows that these frequencies also vary depending on the bearing manufacturer and that

RGR can have a bearing frequency coefficient of 0.378;

Βδ may have a bearing frequency coefficient of 1.928;

- 13 021908

OK can have a bearing frequency of 3.024 and

ΙΚ may have a bearing frequency coefficient of 4.976.

The frequency coefficient is multiplied by the speed of rotation of the shaft to obtain a repetition rate. The work indicates that for a shaft having a rotation speed of 1715 rpm, i.e. 28.58 Hz, the repetition rate for a pulse emanating from the inner ring (OK) of a standard type 6311 bearing may be approximately 86 Hz; the repetition rate of the PTP can be 10.8 Hz.

When the tracked shaft rotates at a constant speed, this repetition rate can be explained either in repetition units per time unit or in repetition units per revolution of the tracked shaft without distinguishing between the above two units. However, if a part of the machine rotates at a variable rotational speed, the situation is further complicated, as explained below in connection with FIG. 16, 17 and 20.

Machinery representing sudden damage.

Some types of machinery can experience complete damage or breakdown of the machine very quickly. For some types of machines, for example, rotating parts in a wind farm, there are cases when a breakdown occurs suddenly and as a complete surprise for maintenance and repair specialists and the owner of the machine. Such a sudden breakdown leads to significant costs for the owner of the machine and can lead to other negative side effects, for example, if parts of the machine fall off as a result of an unexpected mechanical malfunction.

The inventor has found that, in particular, there is a high level of noise in the mechanical vibrations of certain machinery and that such noise levels prevent the detection of damage to machines. Therefore, for some types of machinery, traditional methods for prophylactic condition monitoring do not allow a sufficiently early and / or reliable warning of an approaching deterioration. The inventor has concluded that there may be mechanical vibration_mand indicating a deteriorated condition in such machinery, but traditional methods for measuring vibrations to date may not be suitable.

The inventor also found that machines having slowly rotating parts are among types of machinery that are more likely to be subject to sudden failures.

Having realized that, in particular, the high noise level in the mechanical vibrations of certain machinery prevents the detection of damage to machines, the inventor has developed a method for providing the possibility of detecting weak mechanical signals in a noisy environment. As mentioned above, frequency G_from repetition of the component of the vibration signal in the measuring signal §_ea depends on mechanical vibration_mand, which serves as a sign of incipient damage to the rotating part 8 of the tracked machine 6. The inventor has found that it is possible to detect incipient damage, i.e. damage that is only beginning to develop if the corresponding weak signal can vary.

Therefore, the measurement signal may include at least one component_E a vibration signal depending on the vibrational movement of the rotationally moving part 8; the component of the vibration signal has a frequency of_e repetition depending on speed G_Cat rotation of the rotationally moving part 8. The presence of a vibration signal component, which depends on the vibrational movement of the rotationally moving part 8, can therefore provide an early indicator of a deteriorating condition or incipient damage of the machine being tracked.

In an application in air turbines, a shaft whose bearing is being analyzed can rotate at a speed of less than 120 revolutions per minute (rpm), i.e. frequency g_cat shaft rotation is less than 2 revolutions per second (r / s). Sometimes such a shaft, which must be analyzed, rotates at a speed of less than 50 rpm, i.e. with frequency G_cat shaft rotation less than 0.83 r / s. In fact, the rotation speed may typically be less than 15 rpm. While a shaft having a rotation speed of 1715 rpm, as explained in the above work, performs 500 revolutions in just 17.5 s, it takes 10 minutes for a shaft rotating at 50 rpm to complete 500 revolutions. Certain large wind farms have shafts that can typically rotate at 12 rpm = 0.2 rpm.

Accordingly, when the bearing to be analyzed is coupled to a slowly rotating shaft, and the bearing is tracked by a detector generating an analog measurement signal 8_eawhich is sampled using frequency G₈ sampling at approximately 100 kHz, the number of sampled values ​​associated with one full revolution of the shaft becomes very large. As an illustrative example, 60 million (60,000,000) sample values ​​are required at a sampling frequency of 100 kHz to describe 500 revolutions when the shaft rotates at 50 rpm.

In addition, performing advanced mathematical analysis of the signal requires a considerable amount of time when the signal includes so many samples. Accordingly, it is required to reduce the number

- 14 021908 samples per second before subsequent processing of the signal δ_ΕΝν.

FIG. 9 illustrates another embodiment of a preprocessor 200. An embodiment of FIG. 9, the preprocessor 200 includes a digital band-pass filter 240 and a digital envelope shaping module 250, as described above in connection with FIG. 6. As mentioned above, the signal δ_ΕΝν is a digital representation of the envelope signal, which is formed in response to the filtered signal δ_Ε measurement data.

According to the embodiment of FIG. 9 preprocessor 200 digital signal δ_ΕΝν the envelope is delivered to the decimator 310, configured to generate a digital signal δ ^ Εϋ having a reduced frequency G_g] discretization. The decimator 310 is configured to generate a digital output signal, wherein the time duration between two consecutive sampled values ​​exceeds the time duration between two consecutive sampled values ​​in the input signal. The decimator is described in more detail in connection with FIG. 14 below. According to an embodiment of the invention, an optional low-pass filter 280 may be omitted, as mentioned above. When, in the embodiment of FIG. 9, a signal generated by a digital rectifier 270 is delivered to a decimator 310, which includes low-pass filtering, a low-pass filter 280 may be omitted.

Decimator 310 output 312 delivers a digital signal δ_ΚΕ._υ input 315 of the module 320 increasing the signal-to-noise ratio. The signal-to-noise ratio enhancement module 320 allows the reception of a digital signal δ_ΕΕϋ and in response to this formation of the output signal δ_ΜυΡ. Output signal δ_ΜυΡ delivered to port 260 of the output of preprocessor 200.

FIG. 10A is a flowchart that illustrates embodiments of a method for improving patterns of repeating signals in signals. This method can advantageously be used to improve patterns of repetitive signals in signals representing the state of a machine having a rotating shaft. The signal-to-noise ratio enhancement module 320 may be configured to operate according to the method illustrated by FIG. 10A.

Steps δ1000-δ1040 of the method of FIG. 10A represent preliminary steps to be taken to make adjustments before actually generating output signal values. When the preliminary steps are performed, the output values ​​of the signals can be calculated as described in relation to step δ1050.

FIG. 10B is a flowchart illustrating a method for generating a digital output signal. More specifically, FIG. 10B illustrates an embodiment of a method to generate a digital output signal when the preliminary steps described in relation to steps δ1000-δ1040 in FIG. 10A.

With respect to step δ1000 in FIG. 10A, the required duration of Oi / Lt of the output signal δ ^ ρ is determined.

FIG. 11 is a schematic illustration of a first storage device having several positions ί in a storage device. Positions ί in the storage device of the first storage device store an example input signal I containing a sequence of digital values. An example input signal is used to calculate the output signal δ_ΜΕ; Ί&gt; according to embodiments of the invention. FIG. 11 shows some of a plurality of consecutive digital values ​​for an input signal I. The digital values ​​2080 in the input signal I illustrate only a few digital values ​​that are present in the input signal. In FIG. 11 two adjacent digital values ​​in the input signal are separated by the duration of the D_e14a. Value of C_e11a is the inverse of the frequency G_well sampling the input signal received by the signal to noise ratio enhancement module 320 (see FIGS. 9 and 16).

FIG. 12 is a schematic illustration of a second storage device having multiple positions 1 in a storage device. Position I in the storage device of the second storage device stores an exemplary output signal BMO_Rcontaining a sequence of digital values. Therefore, FIG. 12 illustrates a portion of a storage device having digital values ​​3090 stored in consecutive positions in a storage device. FIG. 12 shows consecutive digital values ​​for the output of the BMO_R. Numeric values ​​3090 in the output of the BMO_R illustrate only a few digital values ​​that are present in the output signal. In FIG. 12 two adjacent digital values ​​in the output signal can be temporarily separated by the duration of the D_e14a.

With respect to step δ1000 in FIG. 10, the required duration Ο_ΕΕΝο_ΤΗ 3010 output signal δΜο_Ρ can be selected so that the output signal δΜο can be used_Ρ to analyze certain frequencies in the output signal. If, for example, low frequencies are of interest, a longer output signal is required than when high frequencies are of interest. The smallest frequency that can be analyzed using the output signal is 1 / (O_EZHOTNxc_e14a), where Ο_ΡΕΝο_ΤΗ is the number of sampled values ​​in the output signal. If r_well is the sampling frequency of the input signal I, then the time C_e11a between each digital sample value is 1 / G_well.

- 15 021908

As mentioned above, patterns of repetitive signals may occur in a data signal representing mechanical vibrations. Accordingly, the measurement signal, for example, the signal §ΕΝν delivered by the envelope generating unit 250, and the signal 3_REand, delivered to the signal-to-noise ratio enhancement module 320, may include at least one component 3_FROM a vibration signal depending on the vibrational movement of the rotationally moving part 8; while component 3_FROM vibration signal has a frequency Γ_υ repetition, which depends on the speed G_MOUTH rotation of the rotationally moving part 8. Therefore, in order to clearly detect the occurrence of a pattern of repeating signals having a repetition frequency G_PEP= R₀= 1 / (Oh_EMRTNx1a_ec_but), the output signal is 8 \_SHR must include at least about_LYRTN digital values ​​when consecutive digital values ​​in the output signal 3_MPP separated by duration k | e |_R|.

According to an embodiment, the user can enter a value representing the lowest frequency G_RERT1P the repetition to be detected, as well as information about the lowest expected shaft speed to be monitored. The analysis system 2 (FIG. 1) includes functionality for calculating a suitable value for the variable O [,_E\ c_T|| in response to these values.

Alternatively, with reference to FIG. 2A, the user of the analysis device 14 may set the value O_LYRTN 3010 output 3_MPP by entering the appropriate value through the user interface 102.

In the next step 31010, a duration coefficient b is selected. The duration coefficient b determines how well stochastic signals are suppressed in the output signal 3mo_R. A larger value of b gives less stochastic signals in the output signal 3mo_R, the smaller the value of b. Therefore, the duration coefficient b can be referred to as the value of the signal-to-noise ratio enhancement module. In one embodiment of the method, b is an integer from 1 to 10, but b can also be set equal to other values. According to an embodiment of the method, the value of b can be pre-set in the signal-to-noise ratio enhancement module 320. According to another embodiment of the method, the value b is entered by the user of the method through the user interface 102 (FIG. 2A). The value of the coefficient b also affects the calculation time required in order to calculate the output signal. A larger value of b requires a longer calculation time than a smaller value of b.

Then, at step 31020, the initial position 3 is set_3ART. Starting position 3_3ART is the position in input I.

Starting position 3_3ART is set so as to prevent or reduce the occurrence of non-repeating patterns in the output signal 3mo_R. When starting position 3_3ART is set so that the portion 2070 of the input signal to the initial position has a duration that corresponds to a certain time interval T_3OTOSNA3T1S__MAX, then stochastic signals with the corresponding frequency T_3OTOSNA3T1S__MAX, and higher frequencies are attenuated in the output signal, 3mo_R.

The next step is calculated 31030 the desired duration of the input data signal. The required duration of the input data signal is calculated in step 31030 according to the formula (1) ΐ 1) 1⁼0net * 1 '⁺5 £ Tact⁺0Bietn

Then, in step 31040, the duration C is calculated_LYRTN in the input data signal. Duration C_RE\_1РР| ι is the duration during which the calculation of the output data signal is performed. This duration is C_RE\-_RTN calculated according to the formula (3) (3) СЕШЗТН⁼one BREST ^- 3 5 TACT “% UNSTN

Formula (3) can also be written as

<img file="EA021908B1_D0003.tif" />

The output signal is then calculated in step 31050. The output signal is calculated according to the following formula (5). In formula (5), the value for the output signal is calculated for the ΐ time value in the output signal.

1=^FROMLYOTN (5) 3Μϋρ (ΐ)⁼Σ I (ΐ) (T + ZdTACT + b) ^{where 1} &lt;1&lt;^ABOUTLEMERT^;

1=1.

3mo output_R has duration oh_Eyou_Nas mentioned above. To detect all 3mo output_R value for each time value from ΐ = 1 to 1 = О_LEMERT must be calculated using formula (5). In FIG. 11, the digital value 2081 illustrates one digital value that is used in calculating the output signal. The digital value 2081 illustrates one digital value that is used in calculating the output signal, where 1 = 1. The digital value 2082 illustrates another digital value that is used in calculating the output signal. Reference number 2082 means the digital value I (1 + 3_3ART+1) in the above formula (5), when 1 = 1 and 1 = 1. Therefore, reference numeral 2082 illustrates the value of digital sampling at position number P in the input signal

P = 1 + 5 5TMT +1⁼ 3 £ T AET + 2

In FIG. 12 reference number 3091 means digital selective value §mo_R(1) in the output, where 1 = 1.

The following describes another variant of the method for working with the module 320 increase the signal-to-noise ratio to improve repetitive patterns in signals representing the state of a machine having a rotating shaft. According to an embodiment, the duration Ορενοτη can be pre-set in the signal to noise ratio enhancement module 320. According to other embodiments of the method, the duration is O_Bj,_from may be defined by user input through user interface 102 (FIG. 2A). According to a preferred embodiment of the method, the variable Ο_εενοτη is set to an even integer, which can be divided into two (2) without providing a fractional number. Variable selection Ο_εενοτηaccording to this rule, it advantageously adapts the number of samples in the output signal so that it is suitable for use in the optional fast Fourier transform module 294. Therefore, according to the options for implementing the method, the variable Ο_εενοτη can preferably be set to a number such as, for example, 1024, 2048, 4096.

In a particular advantageous embodiment, the meaning of §§_TaeT is set, in step 1010, so that the input signal part 2070 to the initial position must have a duration identical to the duration of the output signal 3040, i.e. §§_τΑε_Τ= Ο_εενοτη.

As mentioned in connection with the above equation (1), the required duration of the input data signal is

IEN 5TH-θΤΕΝβΤΗ * b + 3 3-GAR.T + θ b EMSTN

Therefore, the assignment of §§_MELTING= O_EEEOT|| in equation (1) gives

I⁼OYEISTN * L + OYENSTN + OYENSTN — Οΐ-ΕΝΟΤΗ * L + OYENYTN * 2

Accordingly, the required duration of the input signal can be expressed in units of the duration of the output signal according to equation (6) (6}⁼ (B + 2) where b is the duration factor explained above; but

ABOUT_EEAATH is the number of digital values ​​in the output signal, as explained above.

Duration C_EEAAATq can be calculated in this embodiment according to formula (7) (7)⁼B * Orepetn

When the preliminary steps described in relation to steps §1000-§1040 in FIG. 10A are implemented, a digital output signal may be generated by a method as described with reference to FIG. 10B. According to an embodiment of the invention, the method described with reference to FIG. 10B is performed by Ό§Ρ 50 (FIG. 2A).

In step 1100 (FIG. 10B), the signal-to-noise ratio enhancement module 320 receives a digital input signal I having a first set 1_EEAAATq sample values ​​at input 315 (see Fig. 9 and / or 16). As indicated above, the digital input signal I can represent mechanical vibrations resulting from the rotation of the shaft, which leads to a vibration having a period T_erepetition.

The values ​​of the received signal are stored (step 1120) in the storage part of the input signals of the storage device associated with the signal-to-noise ratio enhancement module 320. According to an embodiment of the invention, the storage device may be implemented by means of random access memory 52 (FIG. 2A).

In step 1130, the variable 1 used in the above equation (5) is set equal to the initial value. The initial value may be 1 (unit).

At §1140, the output sample value §mo_R(1) computed for sample number 1. The calculation can use the equation below

8more (1)⁼ £ / (!) * / (V +. &amp; Ί &lt;Η · ί + /) ί · 1

Resulting Selected Value §_Μϋ_Ρ(1) is stored (step §1150, FIG. 10B) in terms of storing the output signals of the memory device 52 (see FIG. 12).

At 1160, the process checks the value of variable 1, and if value 1 represents a number below the required number of output sampled values, O_EEEOTc. step 1160 is performed to increase the value of variable 1 before repeating steps 1140, 1150 and 1160.

If, at step 1160, a value of 1 represents a number equal to the required number of output sampled values, O_EEAAAT|_E step 1180 is performed.

At step §1180, the output signal is O, §_{AE) P} delivered to exit 260 (see FIG. 9 and / or 16).

As mentioned above, a data signal representing mechanical vibrations resulting from shaft rotation may include repetitive signal signatures, and a certain signal signature may thereby be repeated a certain number of times per revolution of the shaft being tracked. In addition, several mutually different signatures of the repeating signals may occur, while mutually different signatures of the repeating signals may have a mutually different repetition rate. A method for improving the signatures of repeating signals in signals, as described above, advantageously provides for the simultaneous detection of multiple signatures of repeating signals having mutually different repetition rates. This advantageously provides a simultaneous determination, for example, of a sign of damage to the inner ring of the bearing and a sign of damage of the outer ring of the bearing in one measurement and analysis session, as described below.

FIG. 13 is a schematic illustration of an exemplary output signal δΜορ containing two signatures 4010 and 4020 of repeating signals. Output signal δΜο_Ρ may contain more signatures of repeating signals than the signals illustrated in FIG. 13, but for illustrative purposes only two signatures of repeating signals are shown. Only some of the plurality of digital values ​​for repeating signal signatures 4010 and 4020 are shown in FIG. 13.

In FIG. 13 illustrates the frequency signal 4020 of the outer ring (OK) and the frequency signal 4010 of the inner ring (1K). As can be seen in FIG. 13, the frequency signal 4020 of the outer ring (OK) has a lower frequency than the frequency signal 4010 of the inner ring (1K). The repetition frequency for the frequency signal 4020 of the outer ring (OK) and the frequency signal 4010 of the inner ring (1K) is 1 / T_0Krespectively 1 / T_{1 TO}.

In the above-described embodiments of the method of working with the signal-to-noise ratio enhancement module 320, in order to improve the repetitive signal patterns, the repetitive signal patterns are amplified in calculating the output signal in step δ1050. A higher gain of the repetitive signal patterns is achieved if the coefficient b is assigned a larger value in step δ1010 than if b is assigned a lower value. A larger value of b means that a longer input signal Ι_ΕΕΝΐ|_ΤΕ required at step δ1030. Longer input 1_her._yettherefore, leads to higher amplification of patterns of repeating signals in the output signal. Therefore, a longer input signal Ι_ΕΕΝΐ|_ΤΕ provides the effect of better attenuation of stochastic signals relative to patterns of repetitive signals in the output signal.

According to an embodiment of the invention, the integer value Ι_ΕΕΝΐ|_ΤΕ can be selected in response to the desired attenuation of stochastic signals. In this embodiment, a duration coefficient b can be determined depending on the selected integer value Cezhtn.

Now, consider an exemplary embodiment of a method for working with a signal-to-noise ratio enhancement module 320 to improve repetitive signal patterns when the method is used to amplify a repetitive signal pattern at a certain lowest frequency. In order to be able to analyze the pattern of repetitive signals with a certain lowest frequency, a certain duration of the output signal is required.

As mentioned above, using a longer data input signal in calculating the output signal results in the pattern of repeating signals being amplified more than if a shorter data input signal is used. If a specific amplification of the repeating signal pattern is required, therefore, a certain duration of the input signal can be used to achieve this specific amplification of the repeating signal pattern.

To illustrate the above embodiment, consider the following example.

The pattern of repetitive signals with the lowest frequency частотой) repetition is of interest. To ensure the determination of such a repeating signal, it is necessary to generate an output signal that allows the indication of a complete cycle, i.e. it should represent the duration Τ_Ι= 1 / ί_Ι. When consecutive sampled output values ​​are separated by period C_e11a sampling, the minimum number of sampled values ​​in the output signal is

As mentioned above, the gain of the repeating signal increases with the duration of the input signal.

As mentioned above, the method described with reference to the above FIGS. 10-13, is configured to improve repetitive signal signatures in a series of measurement data originating from a rotating shaft. The formulation of the signature of the repeating signals should be understood as the sample values ​​[χ (ΐ), χ (! + Τ), χ (! + 2Τ), ..., χ (1 + ηΤ) |, including the amplitude component having a non-stochastic value amplitude, while the duration T between these sampled values ​​is constant if the shaft rotates at a constant speed of rotation. With reference to FIG. 13 it should be understood that the digital values ​​4010 are the result of amplification of several values ​​of the repeating signal in the input signal Ι (see Fig. 11), while the values ​​of the input signals are separated in time by the duration T_{1 TO}. Therefore, in this case, we can conclude that the signature of the repeated signals refers to damage in the internal

- 18 021908 ring of the bearing unit when the period T_well repetition corresponds to the frequency of passage of the balls in the inner ring. Of course, this involves knowing the diameter of the shaft and the speed of rotation. In addition, when there is such a component of the signal of the signature of the repeating signals, there may be a value x of the component of the repeating signals such that χ (ΐ) has an amplitude similar to the amplitude χ (ΐ + Τ), which has an amplitude similar to the amplitude χ (ΐ + 2Τ ), which has an amplitude similar to that of χ (ΐ + ηΤ), χ, etc. When such a signature of the repeating signals is present in the input signal, it can advantageously be determined using the above method, even when the signature of the repeating signals is so weak as to form an amplitude component smaller than the component of the stochastic signals.

The method described in connection with FIG. 10-13 may be performed by the analysis device 14 when the processor 50 executes the corresponding program code 94, as explained in connection with the above FIG. 4. The data processor 50 may include a central processor for controlling the operation of the analysis device 14, as well as a digital signal processor (Ό8Ρ). Ό8Ρ may be configured to actually execute program code 90 to instruct analysis device 14 to execute program 94 instructing the execution of the process described above in connection with FIG. 10-13. A digital signal processor may, for example, be of the type ТМ8320С6722, manufactured by Тс. \ А5 1п51гитсп (5. Thus, the analysis device 14 can be configured to perform all of the signal processing functions 94, including a filter function 240, an envelope generation function 250, a decimation function 310 and 470, and a signal to noise ratio enhancement function 320.

According to another embodiment of the invention, the signal processing may be jointly performed by the device 14 and the computer 33, as mentioned above. Therefore, the device 14 can receive an analog measurement signal 8_ΕΑ and generate the corresponding digital signal 8_Ml and then deliver the digital signal 8_Ml to the control computer 33, making it possible to perform additional functions 94 of the signal processing in the control room 31.

Thinning sampling rate.

As explained above in connection with FIG. 9, it may be desirable to provide decimator 310 to reduce the sampling rate of the digital signal before delivery to the signal to noise ratio enhancement module 320. Such decimator 310 advantageously reduces the number of samples in the signal to be analyzed, thereby reducing the amount of memory needed to store the signal to be used. Thinning also provides faster processing in the subsequent signal to noise ratio enhancement module 320.

FIG. 14A illustrates the number of sampled values ​​in a signal delivered to the input of decimator 310, and FIG. 14B illustrates output sample values ​​of a corresponding time period. The signal input to the decimator 310 may have a frequency ί₈ discretization. As you can see, the output signal has a reduced frequency £_G1 discretization. The decimator 310 is configured to thin out the digital signal 8_ΕΝν envelope so as to deliver a digital signal 8_ΚΕϋ having a reduced sampling frequency, so that the output sampling frequency is reduced by an integer coefficient M compared to the input frequency ί₈ discretization.

Therefore, the output signal 8_ΚΕϋ includes only every Mth sampled value present in input signal 8_ΕΝν. FIG. 14B illustrates an example where M is 4, but M can be any positive integer. According to an embodiment of the invention, the decimator can operate as described in I8 5633811, the contents of which are incorporated herein by reference.

FIG. 15A illustrates a decimator 310 according to an embodiment of the invention. In embodiment 310A of decimator 310 of FIG. 15A, comb filter 400 filters and decimates the incoming signal at a 16: 1 ratio. Those. the output sampling rate is reduced by the first integer coefficient M1 to sixteen (M1 = 16) compared to the input sampling rate. Filter 401 with a finite impulse response (P1K) receives the output of comb filter 400 and provides a further reduction in sampling rate by a second integer coefficient M2. If the integer coefficient is M2 = 4, the P1K filter 401 provides a 4: 1 reduction in the sampling rate, and therefore, the decimator 310A provides a total decimation of 64: 1.

FIG. 15B illustrates another embodiment of the invention in which an embodiment 310B of decimator 310 includes a low pass filter 402, followed by a sampling module 403. Sample selection module 403 is configured to select each Mth sample from a signal received from low pass filter 402. The resulting signal is 8κ_Εϋΐ has a sampling frequency £_G1=£₈/ M, where ί₈ is the sampling frequency of the received signal 8_ΕΝν. The cutoff frequency of the lowpass filter 402 is controlled by the value of M.

According to one embodiment, the value of M is pre-set to a specific value. According to another embodiment, a value of M may be set.

- 19 021908

The decimator 310 may be set to perform the selected decimation M: 1, where M is a positive integer. The value of M can be taken at port 404 of decimator 310.

The cutoff frequency of the lowpass filter 402 is G_G1/ (OhM) Hertz. Coefficient O can be chosen equal to two (2.0) or a value in excess of two (2.0). According to an embodiment, the value of O is selected equal to a value between 2.5 and 3. This advantageously provides the avoidance of aliasing. The lowpass filter 402 may be implemented by a P1K filter.

A signal delivered by a low-pass filter 402 is delivered to a sampling unit 403. The sampling module receives an M value in one port and a signal from a low-pass filter 402 in another port, and it generates a sequence of sample values ​​in response to these inputs. The sample selection module is configured to select each Mth sample from a signal received from a low-pass filter 402. The resulting signal 3 | &lt;yl has a sampling rate^GL &lt;1 = 1 / MxG₈where ί₃ is the sampling rate of the signal δ_ΕΝν received at port 405 of decimator 310.

A method for compensating for variable shaft speed.

As mentioned above, the signature of the repetitive signals present in the input signal can advantageously be determined using the above method, even when the signature of the repetitive signals is so weak that it generates an amplitude component smaller than the components of the stochastic signals.

However, in certain applications, the shaft rotation speed may vary. The implementation of the method described with reference to FIG. 10-13, using the input measurement sequence, when the shaft rotation speed varies, leads to a deteriorated quality of the resulting output signal δ_ΜϋΡ.

Accordingly, an object of an aspect of the invention is to achieve the same high quality of the resulting unit Υ when the shaft rotation speed varies, for example, and when the shaft rotation speed is constant during the entire measurement sequence.

FIG. 16 illustrates an embodiment of the invention including a decimator 310 and a signal-to-noise ratio enhancement module 320, as described above, and a fractional decimator 470.

According to an embodiment of the invention, while decimator 310 is configured to decimitate a sampling rate of M: 1, where M is an integer, the embodiment of FIG. 16 includes a fractional decimator 470 for decimating the sampling rate by υ / Ν, where and and N are positive integers. Therefore, fractional decimator 470 advantageously provides decimation of the sampling rate by a fractional number. According to an embodiment, the values ​​for υ and N can be selected in the range from 2 to 2000. According to an embodiment, the values ​​for υ and N can be selected in the range from 500 to 1500. According to yet another embodiment, the values ​​for υ and N can be selected in the range from 900 to 1100.

In the embodiment of FIG. 16, the output from decimator 310 is delivered to a selection module 460. The selection module provides a signal selection to be input to the signal to noise ratio enhancement module 320. When state monitoring is performed for the rotating part having a constant rotation speed, the selection module 460 may be set in such a position that it delivers a signal 3 | &lt;_Eand · having a frequency ^GЖ1 sampling, at the input 315 of the module 320 increasing the signal-to-noise ratio, and the fractional decimator 470 can be deactivated. When state monitoring is performed for a rotating part having a variable rotation speed, the fractional decimator 470 can be activated, and the selection module 460 is set in such a position that it delivers a signal δ_ΕΕϋ2having a frequency of_G2 sampling, to the input 315 of the module 320 increasing the signal-to-noise ratio.

The fractional decimator 470 has an input 480. The input 480 can be connected to receive a signal output from the decimator 310. The fractional decimator 470 also has an input 490 for receiving information indicating the rotation speed of the shaft 8.

A speed detector 420 (see FIG. 5) may be provided to deliver a signal indicating speed G_KOt of shaft rotation 8. A speed signal can be received at port 430 of the processing means 180, thereby enabling the processing means 180 to deliver this speed signal to the input 490 of the fractional decimator 470. Speed ​​Г_EoT rotations of shaft 8 may be provided in units of rotations per second, i.e. Hertz (Hz).

FIG. 17 illustrates an embodiment of a fractional decimator 470 providing the ability to change the sampling rate to a fractional number, υ / χ where υ and N are positive integers. This provides very precise control of the G frequency._G2 sampling, which should be delivered to the module 320 increase the signal-to-noise ratio, thereby providing a very good definition of weak signatures of repetitive signals, even when the speed of rotation of the shaft varies.

The speed signal received at the input 490 of the fractional decimator 470 is delivered to a generator of 500 fractional numbers. The fractional number generator 500 generates integer outputs and and N at the outputs 510 and 520, respectively. The output and delivered to the upsampling module 530. Upsample module 530 receives signal 3_CES (see FIG. 16) through an input 480. The upsampling module 530 includes a sampling module 540 for entering υ-1 sample values ​​between each sample value received on port 480. Each such added sample value contains an amplitude value. According to an embodiment, each such added sample value is a zero (0) amplitude.

The resulting signal is delivered to a low-pass filter 550, the cutoff frequency of which is controlled by the value of and delivered by the fractional number generator 500. Low-pass filter cut-off frequency 550 is ί_3Κ2/ (Κχυ) Hertz. Coefficient K can be chosen equal to two (2) or a value in excess of two (2).

The resulting signal is delivered to decimator 560. Decimator 560 includes a low-pass filter 570, the cutoff frequency of which is controlled by the value Ν delivered by the fractional number generator 500. The cutoff frequency of the lowpass filter 570 is Γ_3Κ2/ (ΚχΝ) Hertz. Coefficient K can be chosen equal to two (2) or a value in excess of two (2).

A signal delivered by a low-pass filter 570 is delivered to a sampling module 580. The sampling module receives an N value on one port and a signal from a low-pass filter 570 on another port, and it generates a sequence of sample values ​​in response to these inputs. The sampling module is configured to select each Mth sample from a signal received from a low-pass filter 570. Resulting signal 3_ΚΕϋι has a sampling frequency Γ_3K1= 1 / MhG₈where ί₃ is the sampling rate of the signal δ_ΕΝνreceived at port 480 of decimator 590.

Lowpass filters 550 and 570 may be implemented by PC filters. This predominantly eliminates the need to perform multiplications with zero amplitude values ​​inputted by the sampling unit 540.

FIG. 18 illustrates another embodiment of a fractional decimator 470. The embodiment of FIG. 18 advantageously reduces the amount of computation required to generate signal 3_KEE2.

In the embodiment of FIG. 18, the low-pass filter 570 is excluded, so that a signal delivered by the low-pass filter 550 is delivered directly to the sampling unit 580. When the fractional decimator 470 is implemented by hardware , the embodiment of FIG. 18 advantageously reduces the amount of hardware, thereby reducing production costs.

When the fractional decimator 470 is implemented by software, the embodiment of FIG. 18 advantageously reduces the amount of program code to be executed, thereby reducing processor load and increasing execution speed.

With reference to FIG. 17 and 18, the resulting signal 3_KEE2delivered to the output port of fractional decimator 470 has a sampling frequency ί_3Κ2= υ / Νχ £_3Κ1where £_3K1 is the sampling rate of signal 3_KEEreceived at port 480. The fractional value of υ / Ν depends on the speed control signal received at port 490 of the input. As mentioned above, the speed control signal may be a signal indicating the rotation speed of the shaft 8, which can be delivered by the speed detector 420 (see FIG. 1 and / or 5). The speed detector 420 may be implemented by an encoder providing a pulse signal with an appropriately selected resolution so as to provide the desired accuracy of the speed signal. In one embodiment, encoder 420 delivers a full revolution marker signal once per one revolution of shaft 8. Such a revolution marker signal may be in the form of an electrical pulse having a front that can be accurately determined and indicating a specific rotation position of the monitored shaft 8. According to another embodiment, encoder 420 can deliver a plurality of pulsed signals per revolution of the monitored shaft, so as to provide a determination of changes speeds also within one revolution of a shaft.

According to an embodiment, the fractional number generator 500 controls the values ​​of υ and N so that the reduced frequency £_3K2 sampling is important to provide signal 3_KEE2, the number of samples per shaft 8 rotation is almost constant regardless of changes in shaft 8 speed. Accordingly, the higher the values ​​of υ and Ν, the better the ability of fractional decimator 470 to maintain the number of sample values ​​per shaft 8 rotation equal to a practically constant value .

Fractional thinning, as described with reference to FIG. 17 and 18 can be achieved by performing the corresponding steps of the method, and it can be achieved by a computer program 94 stored in the storage device 60, as described above. The computer program may be executed by Ό3Ρ50. Alternatively, the computer program may be executed by means of a user programmable gate array (PPOA) circuit.

The method described in connection with FIG. 10-13, and decimation, as described with reference to FIG. 17 and 18 may be performed by the analysis device 14 when the processor 50 executes the corresponding program code 94, as explained in connection with the above FIG. 4. The data processor 50 may include a central processor 50 for controlling the operation of the analysis device 14, as well as a digital signal processor (IPM) 50B. PPI 50B may be configured to actually execute program code 90 to instruct analysis device 14 to execute program 94 instructing the process described above in connection with FIG. 10-13. According to another embodiment, the processor 50B is a user programmable gate array (PPOA) circuit.

FIG. 19 illustrates decimator 310 and another embodiment of fractional decimator 470. Decimator 310 receives signal Z_E^ having frequency Г₃ discretization, on port 405, and an integer M on port 404, as described above. The decimator 310 delivers a signal Zeki having a frequency of_Threat sampling, to the output 312, which is connected to the input 480 of the fractional decimator 470A. Output frequency^GS | &lt;1 discretization is

<img file="EA021908B1_D0004.tif" />

where M is an integer.

Fractional decimator 470A receives signal Z_CES1having a frequency ^GZK1 sampling, as a sequence of values ​​Z (]) data, and it delivers the output signal Z_CES2 as another sequence of values ​​of K (e) data to output 590.

Fractional decimator 470A may include a storage device, 604, configured to receive and store data values ​​Z (]), as well as information indicating an appropriate rotational speed G_E0T tracked rotating part. Therefore, the storage device 604 can store each data value Z (]) so that it is associated with a value indicating the speed of rotation of the monitored shaft, while determining the value of signal Z_EA sensor corresponding to the value of Z (]) data.

When generating the output values ​​K (d) of the data, the fractional decimator 470A is configured to read the data values ​​3 (]), as well as information indicating the corresponding rotation speed G_K0T, from the storage device 604.

The data values ​​3 (]) read from the storage device 604 are delivered to the sampling unit 540 for inputting I-1 sample values ​​between each sample value received on port 480. Each such added sample value contains an amplitude value. According to an embodiment, each such added sample value is a zero (0) amplitude.

The resulting signal is delivered to a low-pass filter 550, the cutoff frequency of which is controlled by the value of and delivered by the fractional number generator 500, as described above.

The resulting signal is delivered to a sampling module 580. The sampling module receives an N value on one port and a signal from a low-pass filter 550 on another port, and it generates a sequence of sample values ​​in response to these inputs. The sampling module is configured to select each Ν-th sample from a signal received from a low-pass filter 550. Resulting signal 3_EEE2 has a sampling frequency G_g2= and / ZhG_Threatwhere r_ZE1 is the sampling frequency of signal Z_CESreceived on port 480.

Resulting signal 3_EEE2 Delivered to output port 590.

Therefore, the frequency_G2 discretization for output values ​​K (d) of data below the input frequency G_Threat discretization on coefficient Ό. Ό can be set equal to an arbitrary number greater than 1, and it can be a fractional number. According to preferred embodiments, the coefficient Ό is set equal to values ​​from 1.0 to 20.0. In a preferred embodiment, the coefficient Ό is a fractional number set equal to a value from about 1.3 to about 3.0. Coefficient Ό can be obtained by setting integers and and N equal to suitable values. The coefficient Ό is equal to Ν divided by and ϋ = Ν / υ

According to an embodiment of the invention, the integers and and N are set equal to large integers in order to enable the coefficient Ό = Ν / ϋ to follow the speed changes with a minimum of inaccuracy. The choice of variables and and N as integers in excess of 1000 provides predominantly high accuracy in adapting the output sampling frequency to tracking changes in the speed of rotation of the monitored shaft. So, for example, setting N = 500 and u = 1001 provides Ό = 2.002.

The variable Ό is set equal to the appropriate value at the beginning of the measurement, and this value is associated with a certain rotation speed of the rotating part, which should be monitored. After that, during the state monitoring session, the fractional value Ό is automatically adjusted in response to the speed of rotation of the rotating part, which should be monitored, so that the signal output to port 590 provides an almost constant number of sample values ​​per revolution of the monitored rotating parts.

As mentioned above, encoder 420 may deliver a full revolution marker signal once per full revolution of shaft 8. Such a full revolution marker signal may take the form of an electrical impulse having a front that can be accurately detected and indicating a specific rotation position of the monitored shaft 8. A full revolution marker signal, which may be referred to as an index pulse, may be generated at the output of encoder 420 in response to determining a zero-angle pattern on the coding disk that rotates when the tracked shaft rotates. This can be achieved in several ways, as is known to those skilled in the art. The coding disk may, for example, comprise a zero-angle pattern that generates a zero-angle signal with each revolution of the disk. Changes in speed can be detected, for example, by registering a full revolution marker in the memory 604 each time a monitored shaft transmits a certain rotation position, and by associating a full revolution marker with a sample value § (]) received at that moment. Thus, the storage device 604 stores a larger number of samples between two consecutive full revolution markers when the shaft rotates more slowly, since the analog-to-digital converter delivers a constant number of samples of G§ per second.

FIG. 20 is a block diagram of a decimator 310 and another other embodiment of a fractional decimator 470. This embodiment of a fractional decimator is designated 470B. Fractional decimator 470B may include a storage device 604 configured to receive and store data values ​​§ (]), as well as information indicating an appropriate rotational speed G_CAT tracked rotating part. Therefore, the storage device 604 can store each data value § (]) so that it is associated with a value indicating the speed of rotation of the monitored shaft during the determination of the signal value §_ea sensor corresponding to the value of § (]) data.

470V fractional decimator receives signal §_KEE1having a frequency ^gzk1 sampling, as a sequence of values ​​§ (]) of data, and it delivers the output signal §_KEE2having a frequency Г§_K2 discretization, as another sequence of values ​​of K (q) data to output 590.

Fractional decimator 470B may include a storage device 604 configured to receive and store data values ​​§ (]), as well as information indicating an appropriate rotational speed G_CAT tracked rotating part. A storage device 604 may store data values ​​§ (]) in blocks so that each block is associated with a value indicating a relevant rotation speed of the shaft being monitored, as described below in connection with FIG. 21.

The fractional decimator 470B may also include a fractional decimation variable generator 606, which is configured to generate a fractional value Ό. The fractional value Ό may be a floating point number. Therefore, the fractional number can be controlled equal to the value of the floating point number in response to the received value of G_CAT speed, so the value of the floating-point number serves as a sign of the value of G_CAT speeds with a certain inaccuracy. When implemented by a properly programmed Ό§Ρ, as mentioned above, the inaccuracy of the floating point value may depend on the ability of ΌδΡ to generate the floating point value.

In addition, the fractional decimator 470B may also include a PC filter 608. The PC filter 608 is a PC low-pass filter having a certain low-frequency cutoff frequency configured to be punctured by a factor Ό_ΜΑΧ. Coefficient Ό_ΜΑΧ may be set to a suitable value, for example 20,000. In addition, fractional decimator 470B may also include a filter parameter generator 610.

The operation of the fractional decimator 470B is described with reference to FIG. 21 and 22 below.

FIG. 21 is a flowchart illustrating an embodiment of a method of operating with decimator 310 and fractional decimator 470B of FIG. twenty.

At the first stage of §2000, the speed Г_CAT the rotation of the part to be monitored for status is recorded in the storage device 604 (FIGS. 20 and 21), and this can be done almost simultaneously with how the measurement of vibrations or shock pulses begins. According to another embodiment, the rotation speed of the part to be monitored for a condition is observed for a certain period of time. Highest defined speed P_CATmax and the smallest specific speed P_KOTtsh can be recorded, for example, in a storage device 604 (FIGS. 20 and 21).

In step §2010, the recorded speed values ​​are analyzed to determine whether or not the rotation speed varies. If the speed is determined to be constant, the selection module 460 (FIG. 16) can be automatically set in such a position that it delivers a signal §_keehaving a frequency Г§_K1 sampling, to the input 315 of the module 320 increasing the signal-to-noise ratio, and the fractional decimator 470, 470V can be deactivated. If the speed is defined as a variable, the fractional decimator 470, 470V can be automatically activated, and the selection module 460 is automatically set in such a position that it delivers a signal δ_ΕΕϋ₂having a frequency Γ_δΚ2 sampling, to the input 315 of the module 320 increasing the signal-to-noise ratio.

At step δ2020, the user interface 102, 106 displays the recorded value £_cat speeds or £_E0Ttt Gko_Tspeed max and prompts the user to enter the desired value Ο_νorder. As mentioned above, the frequency Γ_ΚΟΤ shaft rotation is often referred to as order 1. Signals of interest can occur about ten times per shaft revolution (order 10). In addition, it may be interesting to analyze the harmonics of some signals, so it may be interesting to measure up to about 100 or about 500 or even higher. Therefore, the user can enter the number Ο_ν order using user interface 102.

In step δ2030, a suitable output frequency Γ is determined_δΚ2 discretization. According to an embodiment, the output frequency Γ_δΕ2 discretization is set equal to Γ_δ|_&lt;2= ΟΟ_ν/ Γ |_{&lt;οΤιΙιιιι}. where C is a constant having a value greater than 2.0.

Ο_ν is a number indicating the relationship between the rotation speed of the monitored part and the repetition rate of the signal to be analyzed.

Rk_0tt is the lowest rotation speed of the monitored part expected during the upcoming measurement session. According to a variant implementation, the value of Eotst is the lowest rotation speed determined in step δ2020, as described above.

The constant C can be chosen equal to a value of 2.00 (two) or higher, taking into account the discrete representation theorem. According to embodiments of the invention, the constant C can be pre-set to a value between 2.40 and 2.70, where k is a coefficient having a value in excess of 2.0.

Accordingly, the coefficient k can be selected equal to a value in excess of 2.0. According to an embodiment, the coefficient C is advantageously selected such that 100 / C / 2 provides an integer. According to an embodiment, coefficient C may be set to 2.56. Choosing C equal to 2.56 provides 100 / C = 256 = 2 to the power of 8.

At step δ2040, the integer value M is selected depending on the detected speed Γ |_&lt;οΤ rotation of the part to be tracked. The value of M can be automatically selected depending on the detected rotation speed of the part to be monitored, so that the intermediate reduced frequency Γ_δΚ] sampling rate must exceed the required frequency Γ_δΚ2 discretization of output signals. Reduced frequency value £_G1 sampling is also selected depending on what magnitude of the change in rotation speed is expected during the measurement session. According to an embodiment, the frequency Γ_δ The sampling rate of an analog-to-digital converter can be 102.4 kHz. According to an embodiment, the integer value M may be set equal to a value between 100 and 512 so as to provide intermediate values ​​of a reduced frequency^ΓδΚ1 sampling between 1024 and 100 Hz.

At step δ2050, the value Ό of the fractional decimation variable is determined. When the rotation speed of the part to be monitored for a condition varies, the value Ό of the fractional decimation variable should vary depending on the instantaneous determined speed value.

According to another embodiment of steps δ2040 and δ2050, the integer value M is set such that the intermediate reduced frequency Γ_δΚ] discretization exceeds Γ_δΚ2 (as determined in step δ2030 above) by at least a percentage value equal to the ratio between the highest determined value £_E0Ttah speed divided by the smallest specific value According to this embodiment, the maximum value Ό_ΜΑΧ the fractional decimation variable is set equal to the value Ο ^^ Γκμ ^ / Γεοι-ι ™, and the minimum value Ό_μΙΝ the fractional decimation variable is set to 1.0. After that, an instantaneous measurement in real time of the actual value of Γ_Κ0Τ speed, and the instantaneous fractional value Ό is set accordingly.

Ρι &lt;_οΤ is a value indicating the measured rotation speed of the rotating part to be monitored.

At step δ2060, the actual measurement begins, and the required total measurement duration can be determined. This duration can be determined depending on the degree of attenuation of the stochastic signals required in the signal-to-noise ratio enhancement module. Therefore, the required total measurement duration can be set so that it corresponds (or so that it exceeds) the duration required to obtain the input signal Ι_ΕΕΝο_ΤΗas explained above in connection with FIG. 10A to 13. As mentioned above in connection with FIG. 10A to 13, longer input signal Ις_ΕΝο_ΤΗ provides the effect of better attenuation of stochastic signals relative to patterns of repetitive signals in the output signal.

The total measurement duration can also be determined depending on the required number of revolutions of the monitored part.

When measurement begins, decimator 310 receives a digital signal δ_ΕΝν at the frequency Γ_δ, and

- 24 021908 he delivers a digital signal §_CES1 at a reduced frequency_K1= G§ / M to the input of 480 fractional decimator. Below signal §_CES1 explained in terms of a signal having sampled values ​​§ (]), where _) is an integer.

At step §2070, the data values ​​§ (]) are recorded in the storage device 604 and the association of each data value with the value of G_K0T rotation speed. According to an embodiment of the invention, the value of G_K0T rotational speeds are read and written at frequency Г_QC= 1000 times per second. Frequency G |_{K &lt;} read and write can be set equal to other values ​​depending on how much the speed G varies_K0T tracked rotating part.

In the next step, §2080, the analysis of the recorded values ​​of the rotation speed and the division of the recorded values ​​§ (]) of the data into data blocks depending on the values ​​of the rotation speed. Thus, a certain number of blocks of data values ​​§ (]) can be generated, each block of data values ​​§ (]) being associated with a rotation speed value. The rotation speed value indicates the rotation speed of the monitored part when the § (]) values ​​of the data of this particular block are recorded. Individual data blocks may have mutually different sizes, i.e. individual blocks can store mutually different numbers of values ​​of § (]) data.

If, for example, the monitored rotating part first rotates at a first speed G_K0T1 during the first period of time, and then it changes the speed to rotate with the second speed G_K0T2 during the second, shorter time period, the recorded data values ​​§ (]) can be divided into two data blocks, the first block of data values ​​being associated with the first value G_K0T1 speed, and the second block of data values ​​is associated with the second value of G_K0T2 speed. In this case, the second data block should contain fewer data values ​​than the first data block, since the second time period is shorter.

According to an embodiment, when all the recorded data values ​​§ (]) are divided into blocks and all the blocks are associated with a rotation speed value, the method then proceeds to step §2090.

At step §2090, the first block of data values ​​§ (]) is selected and the fractional decimation value Ό corresponding to the associated value Г is determined_K0T rotation speed. Associating this value Ό fractional decimation with the first block of values ​​§ (]) of the data.

According to an embodiment, when all the blocks are associated with the corresponding fractional decimation value Ό, the method then proceeds to step §2090. Therefore, the value of the value Ό of fractional decimation is adapted depending on the speed Г_K0T.

At §2100, a block of data values ​​§ (]) and an associated fractional decimation value Ό are selected, as described in §2090 above.

At step 2121, a block of output values ​​of K is formed in response to the selected block of input values ​​of § and the associated fractional decimation value Ό. This can be done as described with reference to FIG. 22.

At step 2121, a check is made whether or not the remaining input data values ​​to be processed exist. If there is another block of input data values ​​that must be processed, then repeat step 2121. If there are no remaining blocks of input data values ​​to be processed, the measurement session ends.

FIG. 22A, 22B, and 22C illustrate a flowchart of an embodiment of a method for operating the fractional decimator 470B of FIG. twenty.

At §2200, a block of input values ​​§ (]) of the data and the associated specific value Ό fractional decimation are received. According to an embodiment, the received data is as described in step 2100 of FIG. 21 above. Input data values ​​§ (]) in the received block of input data values ​​§ are associated with a specific value Ό fractional decimation.

At the steps of §2210-§2390, the P1K filter 608 is adapted for the specific fractional decimation value приним received at the step of §2200, and a set of corresponding output values ​​of K (s |) signals is generated. This is described more specifically below.

At §2210, filter settings suitable for a particular fractional decimation value Ό are selected. As mentioned in connection with the foregoing in FIG. 20, the P1K filter 608 is a P1 low-pass filter having a certain low-frequency cutoff frequency adapted for thinning by a factor Ό_ΜΑΧ. Coefficient Ό_ΜΑΧ may be set equal to a suitable value, for example 20. The value of P_TO the filter ratio is set equal to the value depending on the coefficient Ό_ΜΑΧ and the specific value Ό of the fractional decimation adopted at §2200. Step §2210 may be performed by a filter parameter generator 610 (FIG. 20).

In step §2220, a value x of the initial position is selected in the received input data block § (]). It should be noted that the value x of the initial position should not be an integer. The P1K filter 608 has a length Ε.ινϋτίί · and the value x of the initial position should then be selected depending on the length Pb_Ewell_TN filter and P values_TO filter relationship. P value_TO the filter relationship is as specified in §2210 above. According to an embodiment, the value of x at the initial position may be set equal to x: = P_EEEET/R_TO.

At step 2222, the summing value of the filter’s IFIM is prepared and set equal to the initial value, such as, for example, §IM: = 0,0.

At step 2222, the position _) is selected in the received input data adjacent to and preceding the position x. Position _) can be selected as the integer part of x.

At §2250, the selection of the position of Ppo8 in the ΡΙΚ-filter corresponding to the selected position _) in the received input data is performed. The position of Pro8 may be a fractional number. The position of the PPO8 filter, relative to the middle position of the filter, can be defined as

Groe = [(x-d) * g_th] where P_TO is the value of the filter relation.

At step 2222, a check is made whether or not a certain value of the position of the PPO filter is outside the allowable limit values, i.e. points in position outside the filter. If this occurs, proceed to step 2300 below. Otherwise, proceeds to step §2270.

In step §2270, the filter value is calculated by interpolation. It should be noted that adjacent values ​​of the filtering coefficients in the low-pass ниж-filter, in general, have similar numerical values. Therefore, the interpolation value should preferably be accurate. First, the integer value 1Pro8 of the position is calculated:

1Pro8: = integer part of Ppo8.

The value Roy1 of the filter for the position of Pro8 is

Gua1 = (1Groz) + [(1Groz + 1) - (1Groz)] * [Thunder-1 £ roses] where (1Рро8) and (1Рро8 + 1) are the values ​​in the reference filter and the position of the Рро8 filter is the position between these values.

At §2280, the update of the summing value of the §IM filter is calculated in response to the position _) of the signal:

5iM: = ZiM + Rua1 * 3 (d)

At §2290, the signal is moved to another position: Reference.): =.) - 1. After that, go to step §2250.

At 2300, a position is selected.) In the received input, adjacent and following after x. This position.) Can be selected as the integer part x plus 1 (one), i.e. .): = 1 + the integer part of x.

At step 2323, a position in the ΡΙΚ filter corresponding to the selected position is selected.) In the received input. The position of Pro8 may be a fractional number. The position of the PPO8 filter, relative to the middle position of the filter, can be defined as a thunderstorm = [(] -x) * G_E] where P_TO is the value of the filter relation.

At step 2323, a check is made whether or not a certain value of the position of the PPO filter is outside the allowable limit values, i.e. points in a position outside the filter. If this occurs, proceed to step 2300 below. Otherwise, proceeds to step 2330.

At step 2323, the filter value is calculated by interpolation. It should be noted that adjacent values ​​of the filtering coefficients in the low-pass ниж-filter, in general, have similar numerical values. Therefore, the interpolation value should preferably be accurate. First, the integer value of position GRro8 is calculated:

Schro8: = integer part of Рро8.

The filter value for the position of Pro8 is

<img file="EA021908B1_D0005.tif" />

where (Грро8) and (Грро8 + 1) are the values ​​in the reference filter and the position of the Рро8 filter is the position between these values.

At step 2323, the update calculation of the summing value of the §IM filter in response to the position is performed.) Of the signal:

5iM: = ZiM + Gua1 * 5 (d)

At §2350, a move to another position of the signal is performed:

Assignment.): =.) + 1. After that, go to step §2310.

At step 2323, the output value K (]) of the data is delivered. The output data value K (]) can be delivered to the storage device, so that successive output data values ​​are stored in sequential positions in the storage device. The numerical value of the output value K (]) of the data is: Κ (]): = §ϋΜ.

At step 2323, the position value x is updated: χ: = χ + Ό.

At step 2323, the value.) Of the position is updated.): =.) + 1.

At step 2323, a check is made whether or not the required number of output data values ​​is generated. If the required number of output data values ​​has not been generated, then proceeds to step δ2230. If the required number of output data values ​​is generated, then go to step δ2120 in the method described with respect to FIG. 21. In fact, step δ2390 is configured to ensure that a block of output values ​​P (c |) of the signals corresponding to the block of input values ​​δ of data received in step δ2200 is generated, and that when output values ​​K of signals corresponding to input values ​​δ of data are generated , then step δ2120 in FIG. 21.

The method described with reference to FIG. 22 may be implemented as a subroutine of a computer program, and steps δ2100 and δ2110 may be implemented as a main program.

According to another embodiment of the invention, compensation of the variable shaft speed can be achieved by controlling the clock delivered by the clock 190. As mentioned above, a speed detector 420 (see FIG. 5) can be provided to deliver a signal indicative of speed G_K0T rotation of the shaft 8. A speed signal can be received on port 430 of the processing means 180, thereby enabling the processing means 180 to control the clock 190. Accordingly, the processing means 180 may have a port 440 for delivering a control clock signal. Therefore, the processing means 180 may be configured to control the clock frequency in response to a specific speed G_K0T rotation.

As mentioned in connection with FIG. 2B, the sampling frequency of an analog-to-digital converter depends on the clock frequency. Therefore, the device 14 can be configured to control the clock frequency in response to a certain speed G_K0T rotation, so that the number of sample values ​​per revolution of the monitored rotating part is maintained at an almost constant value, even when the rotation speed varies.

According to yet another embodiment of the invention, the functionality of the signal-to-noise ratio enhancement module 320, 94 can be achieved by a method for generating autocorrelation data, as described in υδ 7010445, the contents of which are incorporated herein by reference. In particular, the digital signal processor 50 may include functionality 94 for performing sequential Fourier transform operations for the digitized signals to provide autocorrelation data.

Gear condition monitoring.

It should be noted that embodiments of the invention can also be used to observe, track, and determine the state of gears. Some embodiments provide, in particular, advantages in monitoring planetary gears comprising planetary gears, gears and / or gears. This is described in more detail below. Planetary gears, gears and / or gears may also be referred to as planetary gears, gears and / or gears.

FIG. 23 is a front view illustrating a planetary gear 700. A planetary gear 700 includes at least one or more ring gears 702, 703, 704 rotating around a central gear 701. The ring gears 702, 703, 704 are commonly referred to as planetary gears, and a central gear 701 is commonly referred to as a sun gear. Planetary gearing 700 may also include the use of gears 705 with internal gearing, usually also called ring gear. Planetary gears 702, 703, 704 may contain P teeth 707, sun gear 701 may contain δ teeth 708, and ring gear 705 may contain A teeth 706. And the teeth on the ring gear 705 are configured to engage with the P teeth on planetary gears 702, 703, 704, which, in turn, are also configured to mesh with the δ teeth on the sun gear 701. However, it should be noted that the sun gear 701 is usually larger than planetary gears 702, 703, 704, as a result of which the illustration shown in FIG. 23 should not be construed as limiting in this regard. When there are different sizes on the sun gear 701 and planetary gears 702, 703, 704, the analysis device 14 can also distinguish between certain states of the various shafts and gears of the planetary gear 700, as should be apparent from the following.

In many planetary gears, one of the three basic components, i.e. the sun gear 701, planetary gears 702, 703, 704 or ring gear 705, is stationary. One of the two remaining components can then serve as an input and provide power to the planetary gear 700. The last remaining component can then serve as an output and receive power from the planetary gear 700. The ratio of input rotation to output rotation depends on the number of teeth in each gear and on the component that is fixed.

FIG. 24 is a schematic side view of the planetary gear 700 of FIG. 23 when viewed in the direction of arrow δν in FIG. 23. An exemplary arrangement 800 including a planetary gear 700 may include at least one sensor 10 and at least one analysis device 14 according to the invention, as described above. Layout 800, for example, may

- 27 021908 used as a gearbox for air turbines.

In the embodiment 800, the ring gear 705 is fixed. Rotating shaft 801 has several movable arms or bearings 801A, 801B, 801C arranged to engage planetary gears 702, 703, 704. After providing input rotation 802 for rotary shaft 801, rotary shaft 801 and movable arms 801A, 801B, 801C and planetary gears 702, 703, 704 can serve as an input and provide power to planetary gear 700. The rotary shaft 801 and planetary gears 702, 703, 704 can then rotate relative to the sun gear 701. The sun gear 701, which can be mounted on the rotary shaft 803, can thereby serve as an output and receive power from the planet gear gear 700. This configuration forms increase in gear ratio C = 1 + -.

As an example, the gear ratio C when used as a gearbox in an air turbine can be implemented such that the output rotation is approximately 5-6 times greater than the input rotation. Planetary gears 702, 703, 704 can be mounted, through bearings 7A, 7B and 7C, respectively, on movable arms or bearings 801A, 801B and 801C (as shown in both Figs. 23, 24). Rotating shaft 801 can be mounted in bearings 7Ό. Similarly, a rotating shaft 803 can be mounted in bearings 7E, and a sun gear 701 can be mounted, through bearings 7P, on a rotating shaft 803.

According to one embodiment of the invention, at least one sensor 10 may be coupled to or at a measurement point 12 of the fixed ring gear 705 of the planetary gear 700. The sensor 10 may also be configured to communicate with the analysis device 14. The analysis device 14 may be configured to analyze the state of the planetary gear 700 based on the measurement data or the signal values ​​delivered by the sensor 10, as described above in this document. The analysis device 14 may include an evaluation module 230, as described above.

FIG. 25 illustrates an analog version of an exemplary signal generated by and output by a preprocessor 200 (see FIGS. 5 or 16) in response to signals detected by at least one sensor 10 when the planetary gear 700 is rotated in layout 800. The signal is shown for duration Τ_ΕΕνwhich represents the signal values ​​determined during one revolution of the rotating shaft 801. It should be understood that the signal delivered by the preprocessor 200 to port 260 (see FIGS. 5 and 16) can be delivered to input 220 of evaluation module 230 (see FIG. . 8 or 7).

As can be seen from the signal in FIG. 25, the amplitude or output of the signal for the signal increases as each of the planetary gears 702, 703, 704 passes the measuring point 12 of the sensor 10 in the arrangement 800. These parts of the signal are referred to below as high amplitude regions 702A, 703A, 704A, which may contain peaks 901 with high amplitude. It can also be shown that the total number of peaks 901, 902 in the signal per revolution of the rotating shaft 801, i.e. during the period Τ | &lt;ι,_ν of time, directly correlates with the number of teeth on the ring gear 705. For example, if the number of teeth on the ring gear 705 is A = 73, the total number of peaks in the signal during the period Τ_ΕΕν time is 73; or if the number of teeth on the ring gear 705 is A = 75, the total number of peaks in the signal during the period Τ_ΕΕν time is 75 etc. As shown, this is true as long as there are no errors or failures in gears 702, 703, 704, 705 of arrangement 800.

FIG. 26 illustrates an example of a portion of the high amplitude region 702A of the signal shown in FIG. 25. This part of the signal can be generated when the planetary gear 702 transmits its mechanically closest position to the measurement point 12 and the sensor 10 (see Figs. 23, 24). It should be noted that small periodic disturbances or vibrations 903, which are illustrated in FIG. 26 may occasionally occur. Here, small periodic disturbances 903 are associated with errors, malfunctions, or ruptures in bearings 7A, as shown in FIG. 23, 24, which can be mounted on one of the movable levers 801A. Small periodic disturbances 903 can thereby propagate (or be transferred) from the bearing 7A through the planetary gear 702 of the planetary gear transmission 700 to the ring gear 705, while small periodic disturbances 903 can be detected by the sensor 10, as described above, for example, in connection with FIG. . 1-24. Similarly, errors, failures or ruptures in bearings 7B or 7C mounted on one of the movable levers 801B or 801C can also generate such small periodic disturbances 903 that can be removed by a sensor in the same way as described above. It should also be noted that small periodic disturbances 903 may also occur as a result of errors, failures, or ruptures in bearings 7P that may be mounted on a rotating shaft 803. The determination of these small periodic disturbances in the signal can serve as a sign of the beginning of a deterioration in the quality of bearings 7A, 7B, 7C and / or 7P or indicate that they are at the limit of their active service life. This, for example, may be important because it can help predict when the planetary gear 700 and / or assembly 800 needs maintenance or replacement.

- 28 021908

According to an embodiment of the invention, the state analyzer 290 in the evaluation unit 230 of the analysis device 14 may be configured to detect these small periodic disturbances 903 in the received signal from the sensor 10. This is made possible by the above-described embodiments of the invention. Small periodic disturbances 903 may also be referred to as shock pulses 903 or vibrations 903. According to an embodiment of the invention, the analysis device 14 using the signal-to-noise ratio enhancement module 320, as described above, provides a determination of these shock pulses 903 or vibrations 903 arising from the bearings 7A (or 7B, 7C or 7P) using the sensor 10 mounted on the ring gear 705, as described above. Although the mechanical shock pulse or vibration signal detected by the sensor 10 attached to the ring gear 705 may be weak, the initialization of the signal-to-noise ratio enhancement module 320, as described above, allows monitoring the condition of bearings 7A (or 7B, 7C or 7P) even though a mechanical shock pulse or vibration signal is propagated through one or more of the planetary gears 702, 703 or 704.

As mentioned above and shown in FIG. 7-9, the state analyzer 290 may be configured to perform a suitable analysis by controlling a signal in the time domain or a signal in the frequency domain. However, the determination of small periodic disturbances 903 in the received signal from the sensor 10 is most suitably described in the frequency domain, as shown in FIG. 27.

FIG. 27 illustrates an exemplary frequency spectrum of a signal containing a small periodic disturbance 903, as illustrated in FIG. 26. The frequency spectrum of the signal contains a peak 904 at a frequency that directly correlates with the engagement or engagement of the teeth of planetary gears 702, 703, 704 and ring gear 705. In fact, the frequency of peak 904 in the frequency spectrum is at ΆχΩ. where A is the total number of teeth of the ring gear 705, and Ω is the number of revolutions per second of the rotating shaft 801, when the rotation 802 occurs at a constant speed of rotation.

In addition to the peak 904 in the frequency spectrum, a small periodic disturbance 903, as illustrated in FIG. 26, may form peaks 905, 906 at frequencies ί цент centered around peak 904 in the frequency spectrum. Peaks 905, 906 at frequencies ί], ί₂ may thereby also be referred to as a symmetrical sideband in the region of the central peak 904. According to an exemplary embodiment of the invention, the state analyzer 290 may be configured to determine one or more peaks in the frequency spectrum and thereby configured to detect small periodic disturbances in the signal, received from the sensor 10. It can also be shown that the peaks 905, 906 at the frequencies ίί, ί₂ touch the central peak 904 according to equations 1, 2 / = ^ χΩ) - (7_Βχ / ^) (Unit 1) /, = (LxL) + (LxLu) (Unit 2) where A is the total number of teeth of the ring gear 705;

Ω is the number of revolutions per second of the rotating shaft 801;

C is the repetition rate of the signature of the repeating signals, which can serve as a sign of a deteriorated state and ί₇₀₂ is the number of revolutions per second of planet 702 around its own center.

The repetition rate of the signature of the repeating signals is a sign of this of the rotating parts, which is the source of the signature of the repeating signals. The repetition rate of the signature of the repetitive signals can also be used to distinguish between different types of degraded conditions, as explained above, for example, in connection with FIG. 8. Accordingly, a certain repetition rate of the signature of the repeating signals can serve as a sign of the main group frequency (Ptr), the rotation frequency of the balls (Βδ), the frequency of the outer ring (OK) or the frequency of the inner ring (ΙΚ) touching the bearing 7A, 7B, 7C or 7P in the planetary gear 700 in arrangement 800 of FIG. 24.

Therefore, as described above, a data signal representing mechanical vibrations resulting from the rotation of one or more shafts, for example, a rotating shaft 801 and / or a rotating shaft 803 (see FIGS. 23, 24), may include several repeating signatures signals, and a certain signal signature can thereby be repeated a certain number of times per revolution of one of the monitored shafts. In addition, several mutually different signatures of the repeating signals may occur, while mutually different signatures of the repeating signals may have mutually different repetition frequencies. A method for improving the signatures of repeating signals in signals, as described above, advantageously provides for the simultaneous determination of many signatures of repeating signals having mutually different repetition frequencies. This advantageously provides simultaneous monitoring of several bearings 7A, 7B, 7C, 7P associated with different shafts 801, 803, using a single detector 10. Simultaneous monitoring may also use the fact that the sun gear 701 and planetary gears 702, 703, 704 are typically of various sizes, which further enables a simple determination of which of the bearings 7A, 7B, 7C, 7P in FIG. 23, 24 forms a small periodic disturbance 903 and, thus, which of the bearings 7A, 7B, 7C, 7P in FIG. 23, 24 may need maintenance or replacement. The method for improving the signatures of repetitive signals in the signals, as described above, also mainly allows you to distinguish between, for example, the damage sign of the inner ring of the bearing and the signature of the damage of the outer ring of the bearing in one measurement and analysis session.

A relevant value for Ω representing the rotation speed of planetary gears 702, 703,

704 may be indicated by a sensor 420 (see FIG. 24). The sensor 420 may be configured to generate a signal indicating the rotation of the shaft 803 relative to the ring gear

705, and from this signal the relevant value for Ω can be calculated when the number of teeth of the ring gear 705, planetary gears 702, 703, 704 and sun gears 701 are known.

FIG. 28 illustrates an example of a portion of the example signal shown in FIG. 25. This example part demonstrates another example of an error or failure that the state analyzer 290 can also determine in a manner similar to that described above. If the tooth in one or more of the gears 701, 702, 703, 704, 705 should break or is practically worn out, the state analyzer 290 can be configured to determine if the tooth is broken or worn out, since it also forms a periodic disturbance, i.e. . due to the lack of engagement of the tooth or adhesion of the missing or erased tooth. This can be detected by the state analyzer 290, for example, in the frequency spectrum of a signal received from the sensor 10. It should also be noted that this type of error or malfunction can be determined by the state analyzer 290 in any type of gear and / or gear. The frequency of this type of tooth engagement error or adhesion error in the gear and / or gear transmission is often at a significantly higher frequency than, for example, the frequency G], G₂ in FIG. 27.

FIG. 29 illustrates yet another embodiment of a state analysis system 2 according to an embodiment of the invention. The sensor 10 is physically associated with a machine 6, which may include a gear 700 having several rotating parts (see FIGS. 1 and 29). The gear train of FIG. 29 may be the planetary gear 700 of FIG. 24. The planetary gear 700, for example, can be used as a gearbox for air turbines.

The sensor module 10 may be a shock pulse measurement sensor configured to generate an analog signal 8_ea, including the component of the vibration signal depending on the oscillatory movement of the rotationally moving part in the gear 700. The sensor 10 delivers an analog signal 8_ea in the layout 920 signal processing.

The signal processing arrangement 920 may include sensors interface 40 and data processing means 50. The interface 40 for the sensors includes an analog-to-digital Converter 44 (Fig. 2A, 2B), generating a digital measuring signal δ_Μϋ. An analog-to-digital converter 44 is connected to the data processing means 50 so as to deliver a signal δ_ΜΕ digital measurement data to the data processing means 50.

The data processing means 50 is connected to the user interface 102. The user interface 102 may include user input means 104, enabling the user to provide user input. Such user input may include selecting the desired analytic function 105, 290, 290T, 290P (FIGS. 4, 7, 8) and / or settings for the functions 94, 250, 310, 470, 470A, 470B, 320, 294 of the signal processing (see Fig. 4, 30).

User interface 102 may also include a display 106, as described, for example, in connection with FIG. 2A of FIG. 5.

FIG. 30 is a block diagram illustrating portions of the signal processing arrangement 920 of FIG. 29 together with user interface 102, 104 and display 106.

The interface 40 for sensors contains an input 42 for receiving an analog signal 8_ea from a shock pulse measurement sensor and an analog-to-digital converter 44. A module 43 for bringing the signals to the required parameters (FIG. 2B) may optionally also be provided. A / D converter 44 samples the received analog signal at a specific sampling rate so as to deliver a signal δ_Μϋ digital measurement data having said specific sampling rate.

Frequency g₂ discretization can be set equal

<img file="EA021908B1_D0006.tif" />

where k is a coefficient having a value greater than 2.0.

Accordingly, the coefficient k can be selected equal to a value in excess of 2.0. Preferably, the coefficient k can be chosen equal to a value between 2.0 and 2.9 in order to avoid the effects of aliasing. Choosing a coefficient to an equal value in excess of 2.2 provides a margin of safety with respect to spectral aliasing effects, as mentioned above in this document. The coefficient k can be chosen equal to the value between 2.2 and 2.9 so as to provide the mentioned safety margin while preventing the formation of too many sample values. According to an embodiment, the coefficient k is advantageously selected such that 100xx / 2 provides an integer. According to an embodiment, the coefficient k may be set to 2.56. The choice of equal to 2.56 provides 100hk = 256 = 2 to the power of 8.

According to an embodiment, the frequency G₃ signal sampling 3_{M |}.) the digital measurement data can be set in a fixed way equal to a certain value Γ3, such as, for example, Γ₃= 102.4 kHz.

Therefore, when the frequency₃ discretization is set in a fixed way equal to a certain value of₃frequency G_3EAtah 3EA analogue signal is £ zEteh⁼£ g / c where r_3EAtah is the highest frequency that should be analyzed in a sampled signal.

Therefore, when the frequency₃ discretization is set in a fixed way equal to a certain value Γ₃= 102.4 kHz, and the coefficient k is set equal to 2.56, the maximum frequency G_3EAtahanalog signal 3_EA is £ s_EAt_butx = £₈/ k = 102 40 0 ​​/ 2.5 6 = 4 0 kHz

Signal 3_{M |}.) digital measurement data having a frequency of₃ sampling is received by a filter 240. According to an embodiment, the filter 240 is a high-pass filter having a cutoff frequency G. This embodiment simplifies the design by replacing the band-pass filter described in connection with FIG. 6, onto a high pass filter 240. The cutoff frequency Gc of the high-pass filter 240 is selected approximately equal to the value of the lowest expected value of G_Rmi mechanical resonant frequency of the resonant sensor 10 measuring shock pulses. When the mechanical resonant frequency Cw is in the range of about 30 to 35 kHz, the high-pass filter 240 can be designed to have a lower cutoff frequency G_Bj= 30 kHz. The high-pass filtered signal is then transmitted to a rectifier 270 and to a low-pass filter 280.

According to an embodiment, it should be possible to use sensors 10 having a resonant frequency in the range of about 20 to 35 kHz. To achieve this, the high-pass filter 240 may be designed to have a lower cutoff frequency G_Bj= 20 kHz.

The output from digital filter 240 is delivered to digital envelope shaping module 250.

While analog devices of the prior art use an analog rectifier to generate an envelope signal in response to a measurement signal, which essentially introduces a bias error into the resulting signal, the digital envelope generator module 250 advantageously generates a true rectifier without bias errors. Accordingly, the digital signal 3 ^_ν the envelope should have a good signal-to-noise ratio, since a sensor mechanically resonating at the resonant frequency in the passband of the digital filter 240 leads to a high signal amplitude. In addition, signal processing performed in the digital domain eliminates the addition of noise and eliminates the addition of bias errors.

According to an embodiment of the invention, an optional low-pass filter 280 in the envelope shaping unit 250 may be omitted. In fact, the optional low-pass filter 280 in the envelope shaping unit 250 is eliminated since the decimator 310 includes a low-pass filter function. Therefore, the envelope shaping unit 250 of FIG. thirty effectively comprises a digital rectifier 270, and a signal generated by the digital rectifier 270 is delivered to an integer decimator 310, which includes low pass filtering.

The integer decimator 310 is configured to decimitate a 3 ^ digital signal_ν envelope to deliver digital signal 3_PEPhaving a reduced frequency G_3Pone sampling, so that the output sampling rate is reduced by an integer coefficient M compared to the input frequency G₃ discretization.

The value of M can be set depending on the specific speed G_MOUTH rotation. The decimator 310 may be set to perform the selected decimation M: 1, where M is a positive integer. The value M can be received on port 404 of decimator 310.

Integer decimation is predominantly performed at several stages using low-pass filters with a finite impulse response, with each PR filter being set with the desired degree of decimation. The advantage associated with performing thinning on multiple filters is that only the last filter should have a steep decline. A steep slope PR filter should essentially have many taps, i.e. A steep slope PR filter should be a long filter. The number of PR taps is an indicator regarding:

1) the amount of memory required to implement the filter,

- 31 021908

2) the number of required calculations,

3) the amount of filtration that the filter can perform; in fact, more taps mean more attenuation in the delay band, less ripple, narrower filters, etc. Therefore, the shorter the filter, the faster it can be performed with Ό8Ρ 50. The length of the P1K filter is also proportional to the degree of achievable thinning. Therefore, according to an embodiment of an integer decimator, decimation is performed in more than two steps.

According to a preferred embodiment, integer decimation is performed in four steps: M1, M2, M3 and M4. Full thinning M equals M1xM2xM3xM4. This can be achieved by providing a set of different P1K filters, which can be combined in several combinations to achieve the desired complete decimation of M. According to an embodiment, eight different P1K filters are provided in the comb.

Mostly, the maximum degree of thinning in the last fourth stage is five (M4 = 5), providing a fairly short filter having a total of 201 taps. Thus, P1Kfilters in steps 1, 2 and 3 can have an even smaller number of taps. In fact, this allows the filters in stages 1, 2 and 3 to have 71 taps or less. In order to achieve complete decimation of M = 4000, three P1K filters can be selected that provide decimation of M1 = 10, M2 = 10 and M3 = 10, and a P1K filter that provides decimation of M4 = 4. This provides an output sample rate.^G8K1 = 25.6 when ί₈= 102400 Hz, and a frequency range of 10 Hz. These four P1K filters have 414 taps in total, and the resulting attenuation in the delay band is very good. In fact, if thinning M = 4000 has to be completed in just one step, it should require approximately 160,000 taps in order to achieve equally good attenuation in the delay band.

The output 312 of the integer decimator 310 is connected to the fractional decimator 470 and to the input of the selection module 460. The selection module provides a signal selection to be input to the signal to noise ratio enhancement module 320.

When state monitoring is performed for the rotating part having a constant rotation speed, the selection module 460 can be set in such a position that it delivers a signal Zsl having a frequency ^G8K1 of sampling, to the input 315 of the module 320 increasing the signal-to-noise ratio, and the fractional decimator 470 can be deactivated. When state monitoring is performed for a rotating part having a variable rotation speed, the fractional decimator 470 can be activated, and the selection module 460 is set in such a position that it delivers a signal 8_ΚΕπ₂having a frequency of_G2 sampling, to the input 315 of the module 320 increasing the signal-to-noise ratio.

Fractional decimator 470 may be implemented by fractional decimator 470B, 94, including an adaptable P1K filter 608, as described in connection with FIG. 20-22 and 4.

Fractional decimator 470 connects to deliver decimated signal 8ε_Επ₂having a lower frequency G_G2 sampling, to the selection module 460, so that when the state analyzer is set to monitor a machine with a variable speed, the output from the fractional decimator 470B is delivered to the signal-to-noise ratio enhancement module 320.

The signal-to-noise ratio enhancement module 320, 94 may be implemented as described in connection with FIG. 10A, 10B, 11, 12, and 13 and 4. The measurement signal inputted to the signal-to-noise ratio enhancement module 320 is signal 8_ΚΕπ (see FIG. 30), which is also illustrated in FIG. 11 as having Ι ^ νοίή sampled values. Signal 8_ΚΕπ is also referred to as I and 2060 in the description of FIG. 11. The signal processing module for increasing the signal-to-noise ratio comprises a discrete autocorrelation for a discrete input signal 8_ΕΕϋ. Output O, also called 8_MCPillustrated in FIG. 12 and 13.

Measuring signal δκ_Εϋ₁, 8ε_Επ, which should be introduced into the signal-to-noise ratio enhancement module, may include at least one component 8l of the vibration signal depending on the vibrational movement of said rotationally moving part; wherein the component of the vibration signal has a repetition rate depending on the speed G &lt;_from rotation of the first part. The repetition rate of the 8L component of the signal can be proportional to the speed G &lt;_fromrotation of the tracked rotating part.

Two different signatures of 8Ό1, 8Ό2 damage can have different frequencies £_ϋ1, Gog will continue to improve, i.e. increase your 8ΝΚ, through the signal-to-noise ratio enhancement module. Therefore, the signal-to-noise ratio enhancement module 320 is advantageously configured to improve various signatures 8Ό1, 8Ό2 having mutually different frequencies Г_C1 and ί_ϋ2repetition. Both of the frequencies₀₁ and ί_ϋ2 repetitions are proportional to the speed G &lt;_from rotation of the monitored rotating part, while₀₁ different from ί_ϋ2 (B₁^&lt; &gt; Ch₂) This can be expressed mathematically as follows:

^GL1 =^k1hGCAT ^{and G}.)2=^to2x^G| &lt;o, t.

where k1 and k2 are positive real values; k1 &lt;&gt; k2;

- 32,021,908 k1 is greater than or equal to one (1) and k2 is greater than or equal to one (1).

The signal-to-noise ratio enhancement module delivers a sequence of output signals to the input of the analyzer 290T in the time domain, so that when the user selects through the user interface 102, 104 to perform analysis in the time domain, the analyzer 290T, 105 in the time domain (FIG. 30 and 4) must fulfill the selected function 105 and deliver relevant data to the display 106. An advantage of the signal-to-noise ratio enhancement module 320 is that it delivers an output signal in the time domain. Therefore, state monitoring functions 105, 290T requiring an input signal in the time domain can be set to work directly with the signal values ​​for the output signal illustrated in FIG. 12 and 13.

When the user selects through the user interface 102, 104 to perform analysis in the frequency domain, the signal-to-noise ratio enhancement module must deliver a sequence of output signals to the fast Fourier transform module 294, and the преобразования transform module must deliver the resulting frequency domain data to the analyzer 290P, 105 in the frequency domain (FIGS. 30 and 4). The analyzer 290P, 105 in the frequency domain must perform the selected function 105 and deliver relevant data to the display 106.

In the embodiment shown in FIG. 29 and 30, it is advantageous for the user to simply perform analysis using a signal-to-noise ratio enhancement module and a fractional decimator.

The following is an example of parameter settings.

To perform analysis in the frequency domain, the user can enter the following data through user interface 102, 104:

1) information indicating the highest frequency of repetition of interest. Frequency ί_Ε repetition is the repetition rate of the signature of interest δ_Ε. This information can be entered in the form of a frequency or in the form of a number Ο_νΗιρ|_ι an order indicating the highest repetition rate of the signature of interest δ [; damage;

2) information indicating the desired improvement of the δΝΚ value for the signature δ_Ε repeating signals. This information may be entered in the form of a value b of the ratio improvement module δ отношения. The value b of the ratio improvement module δΝΚ is also explained below and in connection with the above FIG. 10A;

3) information indicating the required frequency resolution of ΡΡΤ 294, when it is required to perform ΡΡΤ of the signal output from the signal-to-noise ratio enhancement module. It can be set as the value in Ζ of the frequency resolution elements. According to an embodiment of the invention, the frequency resolution Ζ is set by selecting one value Ζ from the group of values.

The group of selected values ​​for frequency resolution Ζ may include:

Ζ = 400,

Ζ = 800,

Ζ = 1600,

Ζ = 3200,

Ζ = 6400.

Therefore, although the signal processing is rather complicated, the arrangement 920 is configured to provide a predominantly simple user interface in terms of information requested by the user. When a user enters or selects values ​​for the above three parameters, all other values ​​are automatically set or pre-set in layout 920.

The value b of the module for improving the ratio δ отношения.

The signal to be input to the signal-to-noise ratio enhancement module may include a component of the vibration signal depending on the vibrational movement of the rotationally moving part; the component of the vibration signal has a frequency £_from repetitions depending on speed £_cat rotation of the first part; wherein the measurement signal includes noise as well as a component of the vibration signal, so that the measurement signal has a first signal-to-noise ratio with respect to the component of the vibration signal. The signal-to-noise ratio enhancement module generates an output signal sequence (O) having repeating signal components corresponding to at least one vibration signal component, so that the output signal sequence (O) has a second signal-to-noise ratio value with respect to the vibration signal component. The inventor established by measurements that the second signal-to-noise ratio significantly exceeds the first signal-to-noise ratio when the value b of the ratio improvement module δ отношения is set to one (1).

In addition, the inventor established by measurements that when the value of the b module for improving the δΝΚ ratio increases to b = 4, the resulting δΝΚ value with respect to the component of the vibration signal in the output signal doubles compared to the δΝΚ value associated with B = 1. An increase in the value of b of the module for improving the ratio to b = 10 is considered to provide an improvement in the associated ZIK value by a factor of 3 for the component of the vibration signal in the output signal, compared with the §K value for the identical input signal when b = 1. Therefore, as the value of b increases, the modulus of improving the relationship with b₁= 1 to b₂ the resulting δNΚ value can increase by the square root of b₂.

Additionally, the user can enter a setting so that the arrangement 920 continues to repeat the measurement. The user can set it to repeat the measurement with a specific period T_RM repetition i.e. always start a new measurement when time T_RM gone. T_RM can be set to one week, or one hour, or ten minutes. The value to be selected for this repetition rate depends on the relevant measurement states.

Since the method of the module for increasing the signal-to-noise ratio requires a lot of input data values, i.e. the number of input sample values ​​can be high, and it is suitable for measuring for slowly rotating parts, the duration of the measurement is sometimes quite considerable. Therefore, there is a risk that the user settings for the measurement repetition rate are incompatible with the duration of the measurements. Therefore, one of the steps performed by arrangement 920 immediately after receiving the above user input is to calculate an estimate of the expected duration T_M measurements. Duration T_M makes up

Tm⁼1E11STn / where Ιι, ιχΌτιι is the number of samples in the signal that must be entered into the signal-to-noise ratio enhancement module in order to achieve measurements according to the selected user settings, as specified below, and Г_G2 is as given below.

Layout 920 is also configured to compare duration T_M measurements with a TPM value of the repetition period selected by the user. If the TPM value of the repetition period is less than or approximately identical to the expected duration T_M of measurements, the parameter controller 930 is configured to provide a warning indication through the user interface 102, 106, for example, by means of suitable text on the display. A warning may also include a sound or a flashing light.

According to an embodiment, arrangement 920 is configured to calculate a suggested minimum value for a value of T_RM the repetition period depending on the calculated estimate of the duration T_M measurements.

Based on the above user settings, the parameter processing controller 930 of the signal processing arrangement 920 allows all parameters to be configured for the signal processing functions 94 (FIG. 4), i.e. tuning an integer decimator and tuning a signal-to-noise ratio enhancement module. In addition, the controller 930 parameters allows the installation of all parameters for fractional decimator if necessary. A parameter controller 930 allows a parameter to be set for PPT 294 when a frequency analysis is required.

The following parameter can be pre-set in layout 920 (FIG. 30): frequency ί₃ discretization of the analog-to-digital Converter 40, 44.

The following parameter can be measured: G_EoT.

As mentioned above, the value of G_EoT parameter can be measured and stored in association with the corresponding sample values ​​of the signal δ_ΚΕϋ1, the selective values ​​of which are entered into the fractional decimator 470V.

The following parameters can be automatically set in layout 920: the sampling frequency in the signal output from the signal-to-noise ratio enhancement module 320:

£ sE2⁼С * Оу * £ к0Т where С is a constant having a value exceeding 2.0;

Ο_ν is the order number entered by the user or calculated in response to the highest frequency value to be tracked as selected by the user;

G_CAT is the instantaneous measured rotation speed of the rotating part during the actual state monitoring;

M = the value of the integer decimator for use in decimator 310 is selected from a table including a set of predefined values ​​for complete integer decimation. In order to select the most suitable value of M, the parameter controller 930 (FIG. 30) first calculates rather accurately the value of M-SacG / GzhghGkotshpGottah, where ί₃ and G_G2 are given above and Gk_Ottt / gk_Ottx is a value indicating the relationship between the lowest speed and the highest rotation speed that must be allowed during measurement. Based on the value of M_sa1c, the selector then selects a suitable M value from the list of preset values. This, for example, can be accomplished by selecting the closest M value that is lower than M_Ca1C from the above table.

R_G1= sampling rate to be delivered from integer decimator 310. P_G1 is set equal to G_8I1= R₃/ M.

- 34 021908

Ό is the fractional decimator value for the fractional decimator. Ό can be set equal to_3K| / G_3K2in which £_3Y and G_3K2 are as defined above.

Oooh real⁼C * 2 where C is a constant having a value greater than 2.0, for example 2.56, as mentioned above;

Ζ is the selected number of frequency resolution elements, i.e. information indicating the desired frequency resolution in PPT 294 when it is required to perform a PPT signal output from the signal-to-noise ratio enhancement module.

3_3ΊΑΕΊ= Ο_ΕΕΝΟΊ·_Η, or a value greater than Ορ ^ οίή, where Ο_εενοίη is as defined immediately above.

I ye νοτ n⁼Oenotn * b⁺3 Zgakt + Ogeist n SJEISTN⁼ I LENBTN ”5 5 ΤΑΚΤ Οχ, ΕΝΟΤΗ

3_Mabout_R(1) = sample values ​​of the output signal as specified in equation (5) (see FIG. 10A).

Therefore, the parameter controller 930 is configured to generate corresponding setpoints as defined above and deliver them to the relevant signal processing functions 94 (FIGS. 30 and 4).

After the output signal is generated by the signal-to-noise ratio enhancement module 320, the state analyzer 290 can be controlled to perform the selected state analysis function 105, 290, 290T, 290P by means of a selection signal delivered to the control signal input 300 (Fig. 30). A selection signal delivered to control signal input 300 may be generated by user interaction with user interface 102 (see FIGS. 2A and 30). When the selected analytic function includes a fast Fourier transform, the analyzer 290P is defined by the selection signal 300 so that it controls the input signal in the frequency domain.

The PPT transform module 294 may be configured to perform fast Fourier transform for a received input signal having a certain number of sample values. An advantageous situation is when a certain number of sample values ​​is set equal to an even integer, which can be divided into two (2) without providing a fractional number.

According to an advantageous embodiment of the invention, the number Ο_ΕΕΝο_ΊΗ samples in the output signal from the signal-to-noise ratio enhancement module are set depending on the frequency resolution Ζ. The relationship between the frequency resolution Ζ and the number Ο_ΕΕΝο_ΊΗ samples in the output signal from the module to increase the signal-to-noise ratio is ^Ο^ ΕNΟΊΗ =^kxΖ, where Ο_ΕΕΝο_ΊΗ is the number of samples of sample values ​​in the signal delivered from the signal-to-noise ratio enhancement module 320;

k is a coefficient having a value in excess of 2.0.

Preferably, the coefficient k can be chosen equal to a value between 2.0 and 2.9 in order to provide a good safety margin while avoiding the formation of too many sample values.

According to an embodiment, the coefficient k is advantageously selected such that 100xx / 2 provides an integer. This selection displays the values ​​for Ο_ςενοίηwhich are adapted as input to the PPT transform module 294. According to an embodiment, the coefficient k can be set to 2.56. The choice of k equal to 2.56 displays 100hk = 256 = 2 to the power of 8.

The table shows examples of user-selected frequency resolution values ​​и and corresponding values ​​for Ο_εενοίη.

<td>to</td><td>Ζ</td><td>Oekotn</td>

<td> 2,56</td><td> 400</td><td> 1024</td>

<td> 2,56</td><td> 800</td><td> 2048</td>

<td> 2, 56</td><td> 1600</td><td> 4096</td>

<td> 2,56</td><td> 3200</td><td> 8192</td>

<td> 2,56</td><td> 6400</td><td> 16384</td>

<td> 2,56</td><td> 12800</td><td> 32768</td>

<td> 2, 56</td><td> 25600</td><td> 65536</td>

<td> 2,56</td><td> 51200</td><td> 131072</td>

- 35 021908