Automatic backlash estimation

(54) Title: Automatic estimation of glitches (56) Published publications: US Al 2005 0 240 364 (47) Abstract:

A method is described for automatically estimating the gap in a control system comprising a controller (C) and a process (P) to be controlled, said controller being able to perform integral control of said process by one or more regulator parameters (K, Ti ) to control a process output (y) of said process with respect to a set value (Y_AP) of this. The method involves monitoring said process output signal (y) to determine a value (dy) that reflects instantaneous differences between setpoint value (ysp) and actual value of said process output (y) over a time interval (zlt); and estimating the gap based on said determined value (dy) and said one or more regulator parameters (X, Ti).

<img file="SE530380C2_D0001.tif" />

530 380

SUMMARY

A method is described for automatically estimating the flaw in a control system comprising a controller (C) and a process (P) to be controlled, said controller being able to perform integral control of said process by one or more controller parameters (K, Ti) to control a process output (y) of said process with respect to a set value (Ysp) thereof. The method involves monitoring said process output signal (y) to determine a value (Jy) that reflects instantaneous differences between setpoint values ​​(y_sp) and fair value of said process output (y) over a time interval (At); and estimating the gap based on said set value (Ay) and said one or more regulator parameters (K, Ti).

530 380

Field of the Invention

The present invention relates generally to the detection and estimation of gaps in a control system. The invention relates, in particular, to a method for automatically estimating gaps in a control system comprising a controller and a process to be controlled, wherein said controller can perform integral control of said process using one or more controller parameters to control a process output of said process with respect to a set value for this. The invention further relates to an associated apparatus, a control system and a computer program product for such automatic estimation of glitches.

Background of the invention

A control system comprises one or more controllable means by which a process is controlled. A control valve is perhaps the most common example of such a controllable body. Control valves are subject to wear and tear, which is well known to anyone who is involved in industrial regulation. After a certain period of operation, this wear results in friction and slip which deteriorates the steering performance. For this reason, valves have been identified as the main source of loop-level problems in process control.

Valves with a high degree of static friction result in a stick-slip movement that causes the control circuits to oscillate. As the degree of friction increases, so does the gap in the linkage mechanism in the valve positioner and actuator. The failure adds a time delay to the control circuit, which impairs control. Since control circuits in process control applications are often coupled to surrounding control circuits, there is also a risk that the interference caused by gaps in one circuit propagates to other circuits.

530 380

Of course, when the static friction or gap becomes extensive, the valve should be repaired or replaced. However, this cannot normally be done without interrupting the process. For this reason, and for other financial reasons, it is interesting to try to keep the valve running for as long as possible. The static friction and the gap can be compensated by using available methods, provided that the degree or extent of static friction or gap is known for the compensation method used.

Although the problems caused by static friction and slip are severe, they are often not detected by the operators at process control plants. The main reason is that the staff reduction has led to a situation where each operator simply has too many control circuits to monitor. For this reason, research has been conducted very actively over the last decade on procedures for automatic performance of surveillance. The industrial use of such procedures has also increased rapidly in recent years.

Regarding static friction, a number of automatic methods are known for detecting static friction control circuits. In the case of gaps, on the other hand, no efficient automatic detection or estimation method is known in the prior art. Instead, a manual examination of the process control operator has been required. An example of such a manual examination according to the prior art will now be briefly described.

Figure 1 shows a block diagram of a gap control circuit, the control circuit comprising a controller C and a process P to be controlled. The controller C receives a setpoint signal y_sp and a process output y as input values ​​and output a control signal u as output. The controller output u is not directly input to the process P; rather, it passes through a gap that gives the true processing input signal u_b. Figure 2 illustrates the function of the gap, where the dead zone 330 is designated d. When the control signal u is reversed, the process signal Ub remains constant until u has passed the dead zone d.

The extent of the gap in the control circuit of Figure 1 can be easily determined manually by the procedure shown in Figure 3. In figure 3, the upper diagram shows the process output y and the lower diagram shows the control signal u. The experiment starts with two step changes in the control signal u in the same direction. If the first step is sufficiently large, the effect of the gap in the second step does not occur. The third step is then done in the opposite direction. Now the control signal u must pass the whole gap before the valve moves. If the last two steps are the same size, the gap is d = Ay / K_P, where Ay is the difference between the process output signals after the second and third steps (see Fig. 3), and K_p is the static process gain (also easily obtained from Fig. 3). This manual procedure can be used every time a control system operator inspects the control circuit.

However, as already explained, there is an ongoing tendency in various industrial areas to reduce staff at process control plants. Consequently, the time intervals between manual inspections of the control circuits can often be long. Therefore, it would have been advantageous to detect and estimate the extent of the gap automatically in different industrial areas, ie without the intervention of any human operator.

Disclosure of the Invention

With reference to the above, an object of the invention is to solve or at least reduce the problems discussed above. More specifically, embodiments of the invention are intended to provide automatic detection and estimation of the degree of gaps in a control circuit. In addition, it is an object to make such an automatic error estimation easy to implement in be530 380 fine, real process control applications. Thus, in the present case, the inventor has realized that such automatic error estimation should be based on normal operating data from the control system, ie no user input in the form of parameters or other conduit should be provided by the control system operator.

In general, the above-mentioned objects are achieved by a method, apparatus, control system and a computer program product according to pending independent claims.

A first aspect of the invention comprises a method for automatically estimating gaps in a control system comprising a regulator and a process to be controlled, wherein said regulator is capable of performing integral control of said process using one or more regulator parameters to control a process output signal. of said process with respect to a set value thereof. The method involves:

monitoring said process output signal to determine a value reflecting instantaneous differences between setpoint and actual value of said process output signal over a time interval; and

- estimating the gap from said determined value and said one or more regulator parameters.

An embodiment of the method involves integrating a control error of said process during said time interval and determining said value as

Ay = J | e | d / 7Ai 't where At = ti₊₁ - ti and e = y_sp - y - [hearing value of said process output] - [actual value of said process output].

In one embodiment of the method, a start time for said time interval is determined by detecting a first zero throughput of said process output with respect to its set value, and an end time of said time interval is determined by detecting a second zero throughput of said process output with respect to its set value.

The gap is advantageously estimated as:

^fK ι^λ-Δί-^K, J

Ay where d is an estimate of a dead zone caused by the gap,

K and Ti are regulator parameters used by said regulator to perform integral control of said process, zlt is said time interval,

Ay is said value which reflects instantaneous differences between setpoint and actual value of said process output, and K_p is a static process amplification of said process.

One embodiment involves setting K_p to a predetermined constant, such as substantially equal to 1.

Typically, the failure is caused by a controllable member such as a mechanical, magnetic, electrical, pneumatic or hydraulic valve in the control system. In such a case, the method may involve verifying that a control signal from said regulator to said controllable means as well as said process output both slowly change signal, wherein said step of estimating the error is performed only upon successful verification.

Specifically, the verification of slowly varying signals can be performed by testing whether said time interval is long compared to a closed time constant of said control system.

530 380

In one embodiment, testing is performed by approximating said closed time constant with a time constant among said one or more regulator parameters and checking whether said time interval exceeds a threshold calculated by multiplying said time constant by a predetermined time constant. In other words: in this embodiment, the test is performed by checking whether & t> N * T_IT where N is a positive number that is chosen approximately for real implementation. Only if the test is successful will one determine or rely on the misappreciation. In one embodiment, N is set to 5.

Further, the method advantageously comprises preventing load interference by finding a maximum absolute difference between setpoint and fair value of said process output during said time interval, verifying that said maximum absolute difference does not exceed a threshold value which is a function of said determined value reflecting instantaneous differences between setpoint and fair value of said process output, and to perform said step of estimating the gap only upon successful verification.

The threshold value can be calculated by multiplying said determined value by a predetermined constant. In this embodiment, therefore, prevention of load disturbances is guaranteed by checking whether the maximum difference e._max &lt;Μ * Δγ, where M is a positive number that is chosen approximately for real implementation, and that one decides or relies on the glitch estimate only if the control is successful. In one embodiment, M is set to 2.

A second aspect of the invention comprises an apparatus for automatically estimating gaps in a control system comprising a controller and a process to be controlled, said controller being able to perform integral control of said process using one or more

530 380 controller parameters for controlling a process output of said process with respect to a set value thereof. The apparatus comprises:

means for monitoring said process output to determine a value reflecting instantaneous differences between setpoint and actual value of said process output over a time interval, and

- means for estimating the gap based on said determined value and said one or more regulator parameters.

Said means for monitoring and said means for estimating are advantageously implemented through a processor of said controller. Alternatively, said means may be implemented by a processing device such as a CPU (Central Processing Unit) or a DSP (Digital Signal Processor / Digital Signal Processor) or any other electronically programmable unit and / or logic unit or a combination of devices in any suitable electronic equipment, including with not limited to a computer of any kind, or one or more integrated circuits.

A third aspect of the invention comprises a control system having a controller and a process to be controlled, said controller being able to perform integral control of said process using one or more controller parameters to control a process output of said process with respect to a set value at this. The control system includes:

means for monitoring said process output to determine a value reflecting instantaneous differences between setpoint and actual value of said process output over a time interval, and

- means for estimating the gap based on said determined value and said one or more regulator parameters.

530 380

A fourth aspect of the invention comprises a computer program product comprising program coding means adapted to perform the steps of any method according to the first aspect when performed by a processor.

The second and fourth aspects may generally have the same purposes and advantages, and the same or directly corresponding features, as the first aspect.

Other objects, features and advantages of the present invention will become apparent from the following detailed description, pending dependent claims and the drawings.

Generally speaking, all terms used in the claims shall be interpreted in accordance with their usual meaning in the technical field, unless otherwise expressly defined herein. All references to an element, a means, a step, etc. (respectively the element, the means, the step, etc.) and a device, a component, etc. (resp. the device, component, etc.) shall be openly interpreted as referring to at least one example of said element, device, component, means, steps, etc., unless otherwise expressly stated. The steps of any method described herein need not be carried out in exactly the order described unless explicitly stated.

Figure Description

Embodiments of the present invention will now be described in greater detail, with reference to the accompanying drawings, in which:

Figure 1 is a block diagram of a control circuit with a gap.

Figure 2 illustrates the function of the gap in the control circuit shown in Figure 1.

Figure 3 illustrates manual gap estimation.

530 380

Figure 4 illustrates the negative inverse of the descriptive function of a gap.

Figure 5 illustrates control of an integrative process with gap Figure 6 illustrates control of a stable process with gap.

Figure 7 illustrates the deterioration in control performance that occurs when controlling a stable process with glitches.

Figure 8 illustrates some of the diagrams shown in Figure 6.

Figure 9 illustrates a flowchart of an automatic slip estimation method according to one embodiment.

Figure 10 illustrates a program framework for the automatic error estimation method shown in Figure 9.

Figure 11 is a schematic illustration of a control system as an example of an environment in which the present invention may be applied.

Figures 12-14 are diagrams of industrial tests performed in the environment of Fig. 11.

Figures 15-17 are diagrams illustrating the effects of gap compensation.

Detailed description of the invention

An embodiment of the present invention will now be described in more detail. Before that, however, a theoretical introduction to the gap should be given. Following the description of the automatic gap estimation in the described embodiment, some industrial tests will be briefly referred to. Finally, gap compensation 530 380 which is advantageously performed based on the estimated gap made available thanks to the present invention will be discussed.

Theoretical introduction to gulp

Again, reference is made to the illustration of the control circuit and dead zone shown in Figures 1 and 2. The descriptive function Y_N of a flaw is:

<img file="SE530380C2_D0002.tif" />

71U \ a) where a is the input amplitude and d is the gap shown in Figure 2. The negative inverse of the descriptive function of the gap is shown in Figure 4 (solid line). Figure 4 also shows the Nyquist curves of two loop transfer functions obtained when the integrative process Pi = e ° '^2s/ (s (1 + 0.8s)) (dashed line) and the stable process P, respectively₂ = l / (l + s)⁴) (dashed line) is controlled by PID controllers.

From this figure, one can conclude that integrative processes controlled by regulators with an integrator result in boundary cycles. On the other hand, stable processes, which are reasonably trimmed, do not yield any boundary cycles.

Since d is divided by a at each position where it occurs (1), the form of the descriptive function is independent of d. This has an interesting consequence. This means that the magnitude d of the gap will affect the oscillation amplitudes, but since the intersection with the Nyquist curve occurs at the same position, the oscillation period will remain the same regardless of the magnitude of the gap.

530 380

Throughout this writing, it is assumed that the controller C in Figure 1 is a PID controller. This is usually the case in practice. However, the results presented herein can be quite easily modified to other regulators having integral effect.

The PID controller used in the following examples has the structure

M

<img file="SE530380C2_D0003.tif" />

(2) where u is the output of the controller, y_3p is the setpoint, y_f is the filtered process output, and the controller parameters are gain K, integral time T ± and derivation time. The regulator has setpoint weights equal to zero in both the proportional and the derivative. This is common with industrial regulators. The process output is filtered through a second-order low-pass filter

<img file="SE530380C2_D0004.tif" />

(3) where Y and Yf ​​are the Laplace transforms of the process output and the filtered process output, respectively. A second order filter is used to guarantee high frequency rolloff in the controller, and the filter time constant is T_f = Td / 5. If a Pi controller is used, it is suggested to use the filter time constant T_f= Ti / lQ.

The following two examples illustrate the problems caused by failures in the feedback circuit.

Example 1 - Controlling an integrative process with glitch

An integrative process with transfer function

<img file="SE530380C2_D0005.tif" />

S '(l + 0.8s) (4)

530 380 is controlled by a PID controller of the mold (2) with the parameters

K = 1.9 Ti = 2.4 T_d = 0,67.

The controller parameters are derived using a design method known as MIGO. A 5% gap (d = 0.05) is introduced into the control circuit.

Figure 4 shows the Nyquist curve of the loop transfer function and the descriptive function of the negative inverse gap. The curves intersect, indicating that a boundary cycle will occur. The analysis of the descriptive function predicts a boundary cycle with an amplitude in the process output of 4.4% and a oscillation period of 7.7 s.

Figure 5 shows the results of the simulations, where a setpoint change is made at t = 0 and a load disturbance is applied at the processing input at t = 100. As shown in the figure, the control circuit oscillates. The amplitude of the process output is 3.2% and the oscillation period is 5.7 s. This is quite close to what was assumed by the analysis of the descriptive function.

Example 2 - Control of a stable process with gap

A transfer function process is controlled by a PID controller of the form (2) with the parameters

K = 1.2 Ti = 2.2 T_A « 1,2.

The controller parameters are derived using the MIGOdesign method. A 5% gap (d = 0.05) is introduced into the control circuit.

530 380 ¹³

Figure 4 shows the Nyquist curve of the loop transfer function and the descriptive function of the negative inverse gap. The curves do not intersect, which indicates that no boundary cycle will occur.

Figure 6 shows the results of the simulations, where a setpoint change is made at t = 0 and a load disturbance is applied at the processing input at t = 100. Furthermore, noise with a standard deviation of 1% is applied to the process output. The figure shows that although there is no boundary cycle as in the previous example, there is a serious deterioration of the control caused by the gap. Due to the noise, the control error will never settle. The control signal must pass the dead zone every time the rate of change of the process input signal is to be changed. This means that there will be no low frequency interference with the process output.

The descriptive function analysis and examples illustrate the control problems that arise when gaps are introduced into the control circuit. Regulatory circuits where integrative processes are controlled with regulators that have integral effect will end up in a boundary cycle swing. These oscillations can be detected by the oscillation detection procedures known per se.

Except for extremely delay-dominant processes, control circuits for stable processes do not normally enter boundary cycle oscillations. However, the steering performance also deteriorates in these cases. This is illustrated in Figure 7. The figure shows how the IAE (Integrated Absolute Error) (left diagram) and the peak error e_max (right diagram) increases in load disturbances caused by failures that occur in the control circuit shown in Example 2 above. The solid line corresponds to a load change of 10%, and the dashed line a load change of 20%.

Both the IAE value and e_max is increased when gap d is reduced, although Smax is very noisy. The increase is dependent on the size of the load disturbances. The results are in good agreement with those previously known.

530 380

The gap introduces a dead time in the regulatory circle. The duration of this dead time is dependent on several conditions and parameters. The dead time occurs only when the control signal action is reversed. The dead time is the time it takes for the control signal to pass the dead zone. A low integral gain K / Ti gives a long dead time. The dead time will be short if the control error is large. This means that the dead time is shorter for large load disturbances than for smaller ones. This explains the results shown in Figure 7.

Stable circuits with glitch are not normally detected by oscillation detection procedures because the oscillation amplitude is quite small.

A new automatic detection and estimation procedure for these processes will now be presented.

Automatic estimation of glitches

A method for automatically estimating gaps according to one embodiment will now be described. The method only deals with stable processes. As mentioned before, integrative processes with failures result in a swinging control circuit, which can be detected by previously known procedures.

The automatic error estimation method generally operates on a control circuit shown in Figure 1. Controller C is a PID controller (or any other type of controller with integral effect) that can be used to generate a control signal

<img file="SE530380C2_D0006.tif" />

edt + KT, - ^d dt (6)

Figure 8 shows part of the simulation given in Figure 6. The process output signal y has been filtered through the filter (3). This means that the process output signal presented in Figure 8 is the signal that introduces the PID algorithm. The signals show it

530 380 typical patterns obtained when a stable process is controlled by a regulator that has integral effect and when there is a gap in the control circuit. The process output signal y is a distance ily from the hearing value while the control signal u drives through the dead zone caused by the gap. When the control signal has changed an amount you move the process output signal to the set point y_sp. The times when the process output y crosses the setpoint are marked at fc = 40 and t = 60 in Figure 8. The time between these no 1 rounds is dt = t_{i + 1} - ti.

The change you make of the control signal is mainly caused by the integral part of the controller C. This means that __t ...

<img file="SE530380C2_D0007.tif" />

<img file="SE530380C2_D0008.tif" />

(7) where (8) see Figure 8.

If the signals change slowly, the process dynamics can be overlooked, and the relationship between the process output y and the control signal u is mainly determined by the static process gain Kp. Specifically, K_p, after transients, the ratio of a change of y to a change of u, the change of the former being caused by a change of the latter. The ratio is thus ty = K_pAu_ch (9) where you_true is the part of you where the gap is over and the valve is moving. This means that

<img file="SE530380C2_D0009.tif" />

(10)

From Equations (7) to (10), the following equation is obtained to estimate the gap:

530 380 d = Δµ - Aw,_rac = γ-ΔγΔί - = 'KP —Δί-Ay (ID

The error estimator (11) assumes that the signals change slowly. An appropriate way to check this is to see if lit is long compared to the closed time constant for the entire control system (including the controller C and the regulated process P). Since the closed time constant for the entire control system is typically not known, the fact that T_LR which is one of the control parameters of the PID controller C and thus known, is closely related to the closed time constant. For this reason, the described embodiment represents a step in verifying that both the control signal u and the process output y slowly change signals by testing whether At> 5Tj. The estimation is performed only if this criterion is met.

The information needed to determine the gap on-line is the controller parameters K and Ti as well as the static process gain K_p. Furthermore, it is necessary to measure Ay from (8), ie to integrate the control error e between the zero passages and the time At between these no passages.

The fact that the process reinforcement K_p used in the algorithm requires some consideration as this amplification is typically unknown. On the other hand, the estimate d is completely insensitive to errors in the estimate of K_p. To see this, (11) is rewritten

at

PKK

Ay (12)

The first step inside the parentheses is always greater than 5 since it is assumed that At> 5T ±. For well-tuned controllers that are applied to processes that are not delay-dominant, the product is KK_P normally greater than 0.5. This means that the first step in (12) dominates and that the error estimator (11) is insensitive to the error of K_p. Since industrial controllers normally operate with standardized signals, K is suitably used_p = 1 as a normal value.

It is important that the disturbance does not cause zero crossings. For this reason, the process output y is not only filtered by the second order filter (3) but an additional second order filter is applied before the signal is processed in the estimation procedure. In the examples presented in this publication, the time constant for this latter filter is Ti / 2.

Online procedures such as this should have a safety net. In the described embodiment, an element of such a safety net is to prevent load interference as such interference can impair the estimation of gaps. To check that the process output has a shape similar to that in Figure 8, only the estimation is performed when e_max &lt;2A_y, where e_maxis the absolute value of the largest control error in the interval [ti in ti + l] ·

Other elements of this safety net may need to be developed during industrial field tests.

Thus, the automatic gap estimation method of this embodiment includes the following steps, which are also shown in the flowchart of Figure 9:

In a first step 100, the necessary variables and counters are initiated. Then, the process output y (filtered as described above) is monitored in step 110 for a first zero pass with respect to the set value y_sp. Once the first zero pass has been detected, execution proceeds to step 120, where the control error e is integrated until a second zero pass has been detected in step 130.

Then, in step 140, the time interval Eat between the first and second zeros is determined. The control described above for slowly varying signals y, u is carried out in

530 380 subsequent steps 150. The execution is stopped if the test fails; otherwise, the execution proceeds to step 160, where Δγ is calculated as explained in (8).

Step 170 is the step for preventing load disturbances as explained above. If the load disturbance is judged to be small enough, ie if e_max &lt;2A_y, the gap estimate d is calculated in step 180; otherwise the implementation will end.

A program framework describing the gap estimator is given in Figure 10.

The functionality of the automatic error estimation method can be performed by any suitable equipment available in actual implementation. As will be seen later with reference to the industrial tests and Figure 11, the functionality can be performed, for example, by appropriate programming of a processor or other logic unit in the hardware equipment implementing the PID controller C (realized by a PID1 controller 32 of the control circuit 30 of Figure 11, which The PID1 controller has a programmable central unit 33 with associated memory 34 capable of storing program code and working data). Alternatively, as again shown in Figure 11, the functionality of the automatic error estimation method can be performed with a conventional computer 50 (such as a personal computer, workstation, laptop or PDA) when appropriately programmed and connected to controller C (PID1 controller 32). of the controlled process P.

The result of the automatic error estimation method, i.e., the error estimation d, can be presented in any suitable way to a control system operator (see 52 in Figure 11), for example, using software running in computer 50, and / or using a user interface of controller C ( PID1 controller 32 in Figure 11). The result can be used in different ways, depending on the wishes of the actual implementation. It can be used as a simple indication

530 380 on detected gap (e.g., when d exceeds a threshold indicating a noticeable gap), or an estimate of the extent of occurrence, or as an alarm trigger to service / maintenance personnel that the controllable body (e.g., valve 38 in Figure 11) is in need of service, repair or replacement, or as input to the compensation step of a gap compensation method performed in conjunction with the gap estimation method. See the last part of this description for more information on gap compensation.

Thus, to conclude the above, a new automatic method (online procedure) for gap estimation has been presented. It is given by Equation (12) and some additional features are summarized in Figures 9 and 10. The estimation method is automatic, which is believed to be favorable for its acceptance in the process control industry. The estimation method can be used in many ways. First, it can be used as a detection procedure in a manner similar to the percussion detection procedures known per se. According to the method, the control performance between the zero crossings is studied. One can conclude that gaps occur in the loop if the frequency in the gaps detections increases.

If gaps are detected, and if the estimated gaps values ​​are close to each other, one can also conclude on the extent of the gaps. This is necessary if the goal is not only to detect glitches but also to compensate for them.

If there is static friction in the control circuit, the gap estimated by the estimation method is the sum of the gap and the dead zone caused by static friction. This is presumed to be a good feature because a gap compensator then not only compensates for the gap but also the static friction.

530 380

In the derivation of the gap estimator above, it was assumed that a PID controller would be used. However, it is an easy matter for one skilled in the art to modify the method to other regulators having integral effect.

Industrial testing

The method for estimating errors has been tested on a flow control circuit in a paper mill. The environment is illustrated schematically in Figure 11, already referred to above. The process part is a pipe 12 where paper pulp is transported from a recovery pulp tower 10 to a tank 60. A PID controller (PID1) 32 controls the pulp flow through a valve 38 and therefore forms a control circuit 30. The process output signal y is the pulp flow, measured in the order of 0-900 m.³/ h of a flow detector (F) 36, and the controller output u is on the order of 0-100%.

Setpoint y_sp is external and is provided by a level controller (PID1) 42 downstream of the tank 60. This means that the flow controller 42 forms a slave control circuit 40 in a cascade configuration. The mass flow is driven by a pump 28 which is controlled by a pressure regulator (PID0) 22 which is included in another control circuit 20 and uses a pressure detector 26.

The flow and pressure regulators interact with each other.

To reduce this interaction, the bandwidth of the flow circuit, which is normally quite fast, has been reduced by introducing a low-pass filter with a time constant of 20 s in the circuit.

A manual test was performed to check the extent of the gap in the valve. The result is shown in Figure 12. The controller output is first increased to ensure that the gap is closed. As the flow increases, the gap closes when the control signal is at the final value u = 39%. The controller output is then switched and reduced in increments of 1%. The first steps do not result in any flow loss, which shows that the control signal is inside the dead zone. However, the step made from the value u = 36% gives a flow reduction, which shows that the gap closes close to this value of the control signal. The test shows that the gap is about d = 3%.

The flow controller is a Pi controller with the parameters K =

0.5 and Τχ = 28 s. The signals used in the controller are normalized to the range [0,1]. The static process gain was estimated to be K_p = 1,3.

Figure 13 shows the result of a recording made during approximately 4000 s. The upper diagram shows the external hearing value (noise signal) and process output signal (flow). The lower diagram shows the control signal. The estimated gap values ​​are shown in the upper diagram. The circuit oscillates due to the oscillating setpoint. The setpoint fluctuations are probably generated by the fluctuations caused by the gap. Figure 13 shows that gap was detected five times during the test, with gap estimates ranging from 2.5% to 3.1%. These estimates are close to those obtained from the manual tests in Figure 12.

Setpoint variations can interfere with the estimator of error. If tested e_max &lt;2Δγ were not present, fifteen detections would have been obtained during the test, and in particular the last two major setpoint changes would have yielded estimates that are too large.

In order to eliminate the disturbances caused by the external setpoint, experiments with a fixed internal setpoint were also performed. Figure 14 shows the results of such a test. The upper diagram shows the constant internal setpoint and process output (flow). The lower diagram shows the control signal. The estimated gap values ​​are shown in the upper diagram.

By comparing Figures 13 and 14, it can be seen that the oscillations caused by the external set point have disappeared.

530 380

Figure 14 shows that there are some low frequency disturbances. They are probably caused by the interaction of the pressure control circuit.

Three gap detections were made during the experiments, with the estimates 1.6%, 2.4% and 2.4% respectively. These values ​​are slightly lower than those obtained in Figure 13. This is expected since the setpoint variations in the previous example amplify the effect of the gap. In addition, the derivation of the method and the simulation examples that have been carried out have already shown that the estimation of errors is expected to be cautious.

In summary, the industrial tests have shown that the gap estimation method also works in an industrial environment with severe low frequency interference. In order to obtain a robust method for a real implementation that is automatic in the sense that no user interaction is needed, it may be necessary to add additional elements of the safety net, which is close at hand for a person skilled in the art.

Compensation for miscarriage

When it is discovered that a control valve has so much error that the control deteriorates, the best course of action can be taken to replace or repair the valve. This is all the more important as the extent of the gap normally increases with time.

However, replacing or repairing a valve usually means that production must be stopped. For this reason, and for the economic reason that it is of interest to use a valve for as long as possible, it is important to compensate for the gap.

A control valve does not normally move by itself or when the control signal is constant, unless the actuator is under-dimensioned or the positioner is unstable. The location of

530 The 380 control signal with respect to the gap is therefore given by the control signal and its history. This means that the gap is an invertible non-linearity.

One way to compensate for the gap is to cause the control signal to jump through the gap each time the control action is reversed. The compensation can be seen as a forward compensation w Upg E Up (13) where u is the controller output, u_FB is the feedback term, for example the output of the PID controller (2), and u_FF is the term that compensates for the gap.

An ideal gap compensation would be

<img file="SE530380C2_D0010.tif" />

you_f dt (14)

This compensation cannot be realized in a noisy environment. One possible modification is to filter the control signal before taking the derivative. It provides the following compensation^UFF - ~^slSⁿ\ dt (15) where u_f is the filtered control signal. Note that the gain of the compensator changes from the actual gap d to a value δ, where δ &lt;d. The filtering of the control signal will introduce a delay in detecting the sign changes in the speed of the control signal. This means that the control signal has already started its way through the gap. For this reason, the compensation must be less than in the ideal case.

There are other opportunities to perform the gap compensation. In (15), the control signal u is passed through a low pass filter for

530 380 to reduce the interference introduced into the controller by the process output signal y. Inside the controller, the measurement signal is fed through a high-pass filter due to the derivation term. First, the noise level is then amplified and then it is reduced by the low-pass filter. A more direct way is to base the feed on the measurement signal directly. One approach that will be used in this writing is

<img file="SE530380C2_D0011.tif" />

(16) where the control error is e = y -y_f and y_f is the filtered process output given by (3). When the control error e changes characters, so does the speed of the integral term in the controller. The feed (16) can therefore be seen as an approach where only the noise-sensitive integral part of the controller is taken into account, and the noise-sensitive proportional and derivative parts are excluded from the compensation.

The gap compensation will now be illustrated for the two examples from a previous section of this publication.

Example 3 - Gap compensation for an integrative process

Let us consider the control problem in Example 1. A gap compensator of the mold (15) is supplied to the controller. The filtered control signal is generated as (1 + 57; / 5)²

C / (_S) (17)

This is a relatively high bandwidth filter. On the other hand, the process output is noise-free in this example. Because of this, the gain of the compensator was chosen equal to the gap, ie δ = d = 0.05, so that the compensator coincides with the ideal compensator (14).

530 380

The results of the simulations are given in Figure 15. When comparing Figures 5 and 15, it is obvious that the gap compensator provides an almost ideal compensation in the noise-free case.

Example 4 - Gap compensation for a stable process

Let us consider the control problem in Example 2. A gap compensator of the mold (16) is supplied to the controller. In this example, compensation is more complicated than in the previous example because the process output is contaminated by noise.

Figure 16 shows the result when a compensator with δ = d = 0.05 is used. It can be seen from the figure that the gain of the compensator is too high and that the compensator causes the loop to swing. A decrease in the compensator gain to δ = 0.4 gives the results shown in Figure 17. This compensator gives a process output that is almost unaffected by the gap. The control signal has some high frequency shifts during certain periods. This could have been avoided by adjusting the filtering of the process output. On the other hand, these variations do not cause valve movements due to the gap.

Figure 7 shows that a gap of 5% gives an increased IAE value of about 45% when the load changes by 20%. With the compensator, this increase is reduced to about 15%.

Conclusions

Slack (and static friction) in control valves is a serious loop-level problem in process control systems. There are two aspects to the problem. First, the non-linearities degrade control performance. A further problem, however, is that the loops that face these problems often remain undetected by staff in process control facilities.

530 380

Methods to detect static friction and to compensate for static friction have been available for a number of years and are used in many industrial plants today. Gap compensation is simple as such, but the known methods are rarely used in process control plants. The main reason for this is that no method of detection of gaps and online estimation of gaps has been presented.

This paper has presented an online method for detecting and estimating gaps in control circuits. The method is automatic in the sense that no information must be provided by the user. The only information needed except for the signals in the control system is regulator parameters. The effectiveness of the method has been demonstrated through simulations and industrial field trials.

The invention has been mainly described above with reference to a few embodiments. However, as will be readily apparent to those skilled in the art, embodiments other than those described above are equally conceivable within the scope of the invention as set forth in the appended claims.

530 380