Root cause diagnostics of aberrations in a controlled process

The present invention relates to industrial process control devices and process control loops. In particular, the present invention relates to such a loop diagnostic device.

Process control loops are used in the process industry to control the operation of processes such as oil refining. A transmitter, which is a typical part of the loop, is installed in the field to measure process variables such as pressure, flow rate, or temperature and transmit them to, for example, control room equipment. Controllers such as valve controllers are also part of the process control loop and control the position of the valve based on control signals received through the control loop or generated internally. Other controllers control, for example, electric motors and solenoids. The control room equipment is also part of the process control loop, where the operator or computer in the control room monitors the process based on process variables received from the transmitter in the field and responds with appropriate control signals. The process can be controlled by sending it to the controller. Another process device that can be part of the control loop is a portable communicator that can monitor and transmit process signals on the process control loop. Generally, these are used to construct a device that forms a loop.

<p> Various techniques have been used to monitor the operation of process control loops and diagnose and identify failures within the loops. However, it is further desired to identify the source or "root cause" of the failure, for example, by identifying individual devices or components in the system that are the cause of the anomaly in process operation. This root cause provides the operator with additional information about which equipment in the process needs repair or replacement.</p>

<p> From various viewpoints, an industrial process diagnostic device capable of identifying an abnormality source, that is, a "root cause" in an industrial process is provided. In one aspect, the diagnostic device contains multiple process formation models, each of which pertains to the physical (ie, actual) implementation of an industrial process. One of the plurality of models is selected and at least one process signal for the selected model and process is used to perform the diagnosis. Based on this diagnosis, the root cause of the abnormality is determined.</p>

The present invention can be used in an industrial process to identify the "root cause" of anomalies that occur in a process. FIG. 1 is a diagram showing an example of an industrial process control system 2 used to control the flow rate of the process fluid system 2, a process pipe 4 carrying the process fluid and a two-wire process control loop flowing a loop current I. Including 6. The transmitter 8, the controller 10, the communicator 12, and the control chamber 14, which are connected to the final control elements in the loop such as actuators, valves, pumps, motors or solenoids, are all part of the process control system 2. It is a department. When anomalies occur during the operation of a process, the present invention is used to identify the cause of the observed anomalies.

Loop 6 is shown in one form for illustration purposes, but is a 4-20mA loop, 2,3 or 4-wire loop, multi-drop loop, HART® or FieldBus or other. Any suitable process control loop can be used, such as a loop that operates according to the digital or analog communication protocol of. During operation, the transmitter 8 senses a process variable such as a flow rate with the sensor 16 and transmits the sensed process variable via the loop 6. Process variables are received by the controller / valve actuator 10, the communicator 12 and / or the control room equipment 14. The controller 10 is shown coupled to the valve 18, which can be adjusted to control the process, thereby changing the flow rate in the pipe 4. The controller 10 receives a control input from, for example, the control chamber 14, the transmitter 8 or the communicator 12 via the loop 6 and adjusts the valve 18 in response. In another embodiment, the controller 10 internally generates a control signal based on the process signal received via the loop 6. The communication device 12 may be the portable communication device shown in FIG. 1, or may be a fixedly installed process unit that monitors the process and executes the calculation. The process apparatus includes, for example, transmitter 8 {3095 type transmitter etc. available from Rosemount Inc.}, controller 10, communicator 12, and control room 14, as shown in FIG. Another type of process unit is coupled to the loop using a PC, a programmable logic unit (PLC), or an appropriate I / O circuit to enable monitoring, management, and / or transmission on the loop. Other computers.

FIG. 2 is a simplified graphic model of the process control loop 50 for controlling the level of liquid in tank 52. As described below, such a model can be selectively used to diagnose the root cause of anomalies during process operation. The level transmitter 54 measures the height of the liquid in the tank 52 and provides a basic process variable (PV) to the controller 56. The controller 56 is shown as a PID controller, but may be any type of controller. The controller 56 also receives a set point (SP) associated with the desired level of liquid in the tank 52. The controller 56 uses a known control algorithm to provide a control request (CD) output to valve 58. The optional valve position sensor 60 can be used to measure the actual position of the valve stem of the valve 58. In this example model, other optional components include a pump 62 configured to draw liquid from the tank 52, a transmitter 64 configured to measure inlet flow velocity, and an outlet flow velocity measurement. Includes the configured transmitter 66. As described below, the model and optional components for the model are stored in memory and can be selected by the operator or other selection method. From a different point of view, the memory can be located and accessible to any device that is coupled to the process or has access to the process signal.

The diagnosis of the present invention on the process control system is preferably performed after the process has been stably run into steady mode. Steady mode is confirmed by observing the mean and standard deviation of the process signal. The mean (μ) and standard deviation (Σ) of each process signal (process variables, control signals, etc.) are calculated for N sets of measurements, and the mean and standard deviation are calculated as follows.

<maths num="1"><img file="JP4635167B2_D0001.tif" /></maths>

<maths num="2"><img file="JP4635167B2_D0002.tif" /></maths>

The number of points N depends on the sampling period and sampling rate of the signal. In Equations 1 and 2, Xi is the value of the process signal taken at sample number i. First, a 10-minute sample period is used with a sample rate of 1 sample per second. In one example, the process average is<u style="single">2540mmH2O (= 100inH2O) (with 25.4mmH2O (= 1inH2O) standard deviation)</u>And the subsequent process average is<u style="single">2463.8mmH2O (= 97inH2O) and 2616.2mmH2O (= 103inH2O)</u>If it is between, the loop is determined to be operating in steady mode. A patent relating to the determination of process stability prior to the initiation of the diagnosis was issued on September 12, 2000 as US Pat. No. 6,119,047, which is hereby incorporated by reference in its entirety.

Once steady operation is reached<u style="single">A pulsed shape that temporarily changes significantly from the average data of other parts</u>Data or spikes should be discarded. One method for identifying such data is to continuously compare the signal means with the signal standard deviation. Two consecutive lines, each providing N sets of measurements<u style="single">each</u>The difference between the mean (μ1 and μ2) of the measured values of the signal block (the block below has similar content) must be less than the standard deviation divided by the square root of the number of samples N. This is expressed as follows.

<maths num="3"><img file="JP4635167B2_D0003.tif" /></maths>

Where μ₁Is the average of the previous block, μ₂Is the average of the current block, N is the number of points in the block, and σ₁Is the standard deviation of the previous block.

Different root causes are identified, depending on the process signals available to perform the diagnosis and used in the model. For example, in the case of the process model shown in Fig. 2, there are three different cases.

<tables num="1"><img file="JP4635167B2_D0004.tif" /></tables>

During the initial training phase, all process signals are collected, for example for 20 minutes. This time can be selected by the user. The mean and standard deviation of these signals are calculated. This training phase is repeated until the process is stable. Once the process is stable, the average of each process signal (μ)_t) And standard deviation (σ_t) Aim value (that is, nominal value) is stored.

In addition, individual process signals are evaluated to ensure that the process is operating properly prior to identifying the root cause of the failure. For example, the basic process variable (PV) is evaluated. In the case of the liquid level illustrated in FIG. 2, it is evaluated as follows.

<tables num="2"><img file="JP4635167B2_D0005.tif" /></tables>

Here, PV_RANGE is a range of liquid levels (range of maximum and minimum values). This value is stored in memory accessible by the process control system when it is formed or is entered by the user. Similarly, the control signal (CD) identifies the next failure.

<tables num="3"><img file="JP4635167B2_D0006.tif" /></tables>

In the example in Table 3, the control request is assumed to be between 0% and 100%. If possible, a similar test is performed on the valve position (VP) process signal.

During the monitoring phase, various process signals are monitored, these are unchanged (NC), upward change (U) (average signal).<u style="single">Obtained during the training phase</u>Greater than average) or downward change (D) (average signal<u style="single">Obtained during the training phase</u>It is judged whether it is smaller than the average). The state of NC is determined by Equation 4.

<maths num="4"><img file="JP4635167B2_D0007.tif" /></maths>

In Equation 4, μt is<u style="single">Obtained during the training phase</u>Average, μ is the average of the current block, N is the number of points in the block, and σt is<u style="single">Obtained during the training phase</u>Standard deviation, μt and σt are training phases, respectively<u style="single">Remembered in</u>Mean and standard deviation. N is the number of samples and μ is the average of the current process signals.

The state of the upward change (U) is specified by Equation 5.

<maths num="5"><img file="JP4635167B2_D0008.tif" /></maths>

Where μt is<u style="single">Obtained during the training phase</u>Average, μ is the average of the current block, N is the number of points in the block, and σt is<u style="single">Obtained during the training phase</u>Standard deviation.

Finally, the state of the downward change (D) is specified by Equation 6.

<maths num="6"><img file="JP4635167B2_D0009.tif" /></maths>

Where μt is<u style="single">Obtained during the training phase</u>Average, μ is the average of the current block, N is the number of points in the block, and σt is<u style="single">Obtained during the training phase</u>Standard deviation.

The number of process signals available can identify different root causes as sources of anomaly within the process. For example, if setpoints, fundamental variables and control request process signals are available, problems related to level sensor drift or valves can be identified. Table 4 shows an example of the rule base.

<tables num="4"><img file="JP4635167B2_D0010.tif" /></tables>

If additional process signals are available, the actual valve position (VP) and root cause can be more specifically identified, as shown in Table 5.

<tables num="5"><img file="JP4635167B2_D0011.tif" /></tables>

Finally, if the inflow rate (IF) and outflow rate (OF) process signals are available, it is also possible to determine if there is a leak in the tank 52 as shown in the rule base of FIG. ..

<tables num="6"><img file="JP4635167B2_D0012.tif" /></tables>

If the change in the process signal does not match any of the rules in Tables 4, 5 and 6, an unknown anomaly output can be output. These rules also apply if process 50 operates based on the pressure difference used to drain the pump 62 or the tank 52.

FIG. 3 is a simplified diagram of the graphic model 100 of the process control loop that controls the flow velocity. This illustrates another example of a process control loop. In FIG. 3, tank 102 (or pump 103 or other differential pressure source) can provide a flow of processing fluid. The transmitter 104 senses the flow velocity and provides the basic process variable (flow velocity) to the controller 106. Controller 106 also receives a set point (SP) and provides a control request (CD) signal to valve 108. The valve 108 can optionally return the actual position (VP) of its valve stem according to the option. Additional options include a pressure transmitter 110 configured to sense process pressure (PT) and a redundant flow transmitter 112 configured to sense redundant flow velocity (FT2).

During operation, the mean and standard deviation are determined in a manner similar to that described for Figure 2 and during the training phase, as described in Equations 1 and 2. However, in general, flow velocity control responds relatively quickly, so it is shorter.<u style="single">Of the training phase</u>A period, for example 2 minutes, is used.

As shown in Table 7, the number of different root causes is identified depending on the number of different process signals available.

<tables num="7"><img file="JP4635167B2_D0013.tif" /></tables>

Prior to identifying the root cause, for example, the rule base in Table 8 is used to check for basic failures.

<tables num="8"><img file="JP4635167B2_D0014.tif" /></tables>

The state of the valve is determined as follows.

<tables num="9"><img file="JP4635167B2_D0015.tif" /></tables>

By using additional process variables, the "root cause" of anomalies within the process is identified. If the set points, basic process variables and control request signals are available, the root cause of the process anomaly can be identified as flow sensor drift or valve trouble:

<tables num="10"><img file="JP4635167B2_D0016.tif" /></tables>

If additional process signals are available, the actual valve position (VP) as well as flow sensor drift or valve trouble can be identified as the root cause:

<tables num="11"><img file="JP4635167B2_D0017.tif" /></tables>

Finally, to measure the second flow velocity variable (FT2), leaks in the process can also be identified if redundant transmitters are used.

<tables num="12"><img file="JP4635167B2_D0018.tif" /></tables>

FIG. 4 is a block diagram showing a process apparatus 100 that implements one embodiment of the present invention. The process apparatus 100 includes a root cause analysis block 102 that receives the control signal CD through the control signal input 104, the process variable PV through the process variable input 106, and the set point SP through the set point input 108. Additional process signal (PS)₁, PS₂...) is received through other inputs depending on the number of additional process signals available, such as process signal inputs 110,111.

The root cause analysis block 102 is also connected to a memory that stores a plurality of process formation models 112. The process formation model 112 is stored, for example, in the system memory. Possible process control in the illustrated embodiment<u style="single">loop</u>There are a total of X different process formation models corresponding to. In this example, each process formation model includes graphic models GM1 ... GMx that provide a graphic representation of the process. It is used to provide a graphic user interface that facilitates the input of configuration data by the operator. For example, the graphic model may be similar to that illustrated in FIGS. 2 and 3.

Each process<u style="single">Formation</u>The model can receive any number of process signals (PS1A, PS1B, ... etc.). In the specific examples shown in Figures 2 and 3, the minimum three process signals needed to identify the root cause of the anomaly in the process are the control request CD, the main process variable PV and the setting point SP. Exists. In certain embodiments<u style="single">Process formation</u>The number of process signals associated with the model is, as required, the least or more of the process signals required to perform the root cause analysis.

Then each<u style="single">Process formation</u>The model can contain any number of optional process signals (OP1A, OP1B, ...). The process signal for each option is input 110,<u style="single">111</u>Corresponds to process signals (PS1, PS2, ...) received through etc. In the example of FIG. 2, the valve position VP, inflow velocity IF and outflow velocity OF are examples of optional process signals. Several<u style="single">Process formation</u>The model can be configured to have no optional process signals.

Each process formation model then finds the root cause based on the received process signals (minimum required process signals PS1A, PS1B ... and some optional process signals OP1A, OP1B, ...). It can contain any number of rulebases used to make decisions.<u style="single">In FIG. 4, the rule base is indicated by RB1A, RBP1B, ....</u>Rule-based examples are shown in Tables 4, 5, 6, 10, 11 and 12 above. It should be noted that the present invention is not limited to the special use of the above rule base to perform root cause analysis. In one aspect, any analytical method can be used, including neural networks, other rule-based, regression learning, fuzzy logic and other known diagnostic methods or methods that have not yet been discovered. In the example shown here, the minimum three process signals received are the control request CD signal, the main process variable PV, and the set point SP signal. However, other signals, fewer signals or combinations of different signals can be used to perform root cause analysis.

Root cause analysis block 102<u style="single">The model selection input 116 is received and one of a plurality of process formation models 112 is selected.</u>The model selection input may be from the operator or another source. Model selection input 116 is continuously used in root cause analysis block 102<u style="single">The only process formation model</u>To identify. Also, in one example,<u style="single">Additional optional process (OP) signals contained in the selected process formation model are used in root cause analysis block 102. The OP signal corresponds to a process signal (PS1, PS2, ...) Actually received through inputs 110, 111, etc.</u>If a graphic user interface is used<u style="single">graphic</u>The model is displayed on the display output 118<u style="single">To.</u>For example, model selection input 116<u style="single">By</u>chosen<u style="single">Process formation</u>The process signals associated with the model (PS1A, PS1B, ...) or the optional process signals (OP1A, OP1B, ...)<u style="single">It is assigned to the process signal in the process formation model. This assignment may be displayed in graphic format.</u>

Once<u style="single">Process formation</u>Once the model is selected,<u style="single">Process formation</u>The process signal used by the model's rulebase is assigned to the actual process signal received from the process. Root cause analysis block 102 performs root cause analysis using any of the required techniques, as described above. Based on the root cause analysis, a root cause output 120 is provided that indicates the root cause of the anomaly of the event that occurred in the process.

FIG. 5 according to an embodiment of the present invention is a simplified block diagram showing a physical implementation of the process apparatus 100. In the example of FIG. 5, the device 100 is connected to the process control loop 132 through the input / output unit 134. Loop 132 is, for example, a two-line loop or other process control loop shown in FIG. Also, the connection does not have to be a direct connection, it may be just a logical connection in which variables from the loop are received through the logical input / output block 134. The microprocessor 136 is connected to the memory 138 and the graphic user interface 140. Memory 138 is shown in FIG.<u style="single">Process formation</u>Used to store variables and program instructions as well as model 112.

Graphic user interface 140<u style="single">Process formation</u>It provides an input for receiving the model selection input 116 as well as the display output 118 of FIG. 4 used during the model selection and formation period. The microprocessor 136 can also connect to an optional database 142 that contains information related to the form and behavior of the process being monitored. For example, many process control or monitoring systems have such a database. One example is the AMS system available from Rosemount Incorporation in Edenpreli, Minnesota.

The root cause process device 100 can be implemented in any process device such as a transmitter, controller, mobile communication device, or control room as shown in FIG. In one embodiment, the process apparatus 100 operates on a computer system or PC located in a control room or other remote location. The process control loop 132 generally consists of several types of fieldbus-based loops, or compound control loops. In such a form, the process apparatus 100 is selected.<u style="single">Process formation</u>The desired process signal can be sent to various devices connected to the control loop for the model. Graphic user interface 140 is shown,<u style="single">Process formation</u>The model may be selected by any selection method and does not need to be selected and formed by human manipulation. For example, appropriate rule-based and model options may be received by device 100 based on configuration information stored elsewhere being provided by other techniques. Instead, the root cause process apparatus 100 may be installed, for example, in the field and placed in the transmitter.

The process variables used here are the basic variables that are generally controlled within the process. The process variables used here mean any variable that represents the state of the process, such as pressure, flow rate, temperature, product level, pH, turbidity, vibration, position, motor current, and other process characteristics. .. A control signal means any signal (other than a process variable) used to control a process. For example, the control signal is the desired process variable value (ie, setting point) regulated by the controller or used to control the process, such as desired temperature, pressure, flow rate, product level, pH, or turbidity. Means. The control signal is a signal supplied to the control element such as a calibration value, an alarm, an alarm state, a valve position signal supplied to the valve actuator, an energy level supplied to the heating element, and a solenoid on / off signal. , Or any other signal related to control of the process. The diagnostic signals used here include information related to the operation of devices and elements within the process control loop. However, it does not include process variables or control signals. For example, diagnostic signals include valve stem position, applied torque or force, actuator pressure, pressure of compressed gas used to drive the valve, voltage, current, power, resistance, capacitance, inductance, device temperature. , Stationion, friction, full-on and full-off positions, stroke, frequency, amplitude, spectrum and spectral components, rigidity, electric or magnetic field strength, duration, strength, operation, motor back electromotive force, motor current, loop-related parameters Includes (control loop resistance, voltage, or current, etc.), or any other parameter detected or measured within the system. Further, a process signal means any signal associated with a process or element within a process, such as a process variable, control signal or diagnostic signal. Process equipment includes any equipment that forms part of a process loop or is connected to a process loop and is used to control or monitor the process.

Although the present invention has been described with reference to preferred embodiments, one of ordinary skill in the art will appreciate that the embodiments and details can be modified without departing from the spirit and scope of the invention. Although two specific processing examples and specific model examples are shown in this specification, the present invention can be adapted to other forms and models generated using known methods or methods discovered in the future. .. Also, other types of rule-based or model forms may be used in the present invention. The present invention can be installed as an independent device and may be a software module added to the software used to control or monitor an industrial process. In one aspect, the invention includes computer instructions and / or storage media used to carry out the invention. The "process model" used here is a logical representation of any process and is not limited to the specific examples described above. The "root cause" is the initial cause (or multiple causes) of a change or anomaly during process operation. Other types of process control loops modeled include, but are not limited to, flow control, level control, temperature control, etc., including regulation and cascade control of gas, liquid, solid or other forms of process material. .. Special examples of loops include, for example, flow control loops with valves driven by differential pressure, level control loops with valves driven by differential pressure, temperature control for flow control, valve pumps. Level adjustment control for driving, flow control with valve driven by pump, level adjustment control for valve cooling condenser, flow adjustment control level adjustment control for cascade supply, liquid temperature adjustment for valve Control, liquid temperature control for flow control, gas flow control with valve driven by differential pressure, gas temperature control for valve, gas pressure control for valve, for flow control Includes gas pressure adjustment control, flow force adjustment control, level adjustment control for cascade reboiler, valve and liquid pressure adjustment control for valve revoir.

<figref num="1">It is a simplified diagram of a process control loop including a transmitter, a controller, a mobile communication device and a control room.</figref><figref num="2">It is a schematic diagram of the process control loop model for a liquid level loop.</figref><figref num="3">It is a schematic diagram of the process control loop model for the flow velocity control loop.</figref><figref num="4">It is a block diagram of the apparatus of one Embodiment of this invention.</figref><figref num="5">It is a block diagram which shows an example of the hardware of FIG.</figref>

Code description

2 ...... Industrial process control system 4 ...... Process pipe 6 ...... 2-wire process control loop 8 ...... Transmitter 10 ...... Controller 12 ...... Communication device 14 ...... Control room 16 ...... Sensor 18 ...... Valve 100 ...... Process equipment 102 ...... Root cause analysis block 104 ...... Control signal input 106 ...... Process variable input 108 ...... Setting point input 110,111 ...... Process signal input 112 ...... Process formation model 116 ...... Model selection input 118 ...... Display output 120 ...... Root cause output