JPH0769722B2 - Self-tuning controller - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a self-tuning controller that automatically adjusts (tunes) at least proportional (P) and integral (I) calculation parameters to optimum values, More specifically, it is configured to observe the response waveform of the deviation signal between the process amount or the process amount and the control target value, and automatically tune the operation parameter to the optimum value based on the evaluation index of the response waveform. It concerns self-tuning controllers.

(Prior Art) A method of observing a response waveform of a process amount or a deviation signal in a controller used for feedback control and automatically tuning an operation parameter to an optimum value based on an evaluation index of the response waveform. There is a controller.

FIG. 6 is a waveform diagram for explaining the operation of the self-tuning controller of such a system disclosed in US Pat. No. 4602.326.

If a response waveform as shown in the figure is observed, the controller
The first, second, and third peak values E ₁ , E ₂ , E ₃ are detected, and the overshoot amount = −E ₂ / E ₁ , tamping = (E ₃ −E ₂ ) / (E ₁ −
E ₂ ), with the time Tp between the first and third peaks as a cycle, the evaluation parameters representative of these response waveforms are tuned so that the evaluation parameters match the evaluation indexes of the predetermined and optimal response model. is doing.

(Problems to be solved by the invention) However, the controller having such an operation calculates the overshoot amount and the damping value by using the peak values E ₁ , E ₂ , and E ₃ of the response waveform, and Is used as an evaluation index. For example, when similar response waveforms as shown in FIGS. 7 (a) and 7 (b) are observed, overshoot of the response waveform is generated even if the magnitude of the amplitude is different. The amount and damping value are the same. Here, in the case of the response waveform shown in FIG. 7 (b), its amplitude is small, and there is a possibility that the amplitude due to noise may be observed as the amplitude of the peak of the response waveform. On the other hand, in the case of the response waveform shown in FIG. 7A, even if noise is mixed in because of its large amplitude, it is not affected as long as it does not overlap the peak of the response waveform.

For this reason, for example, if the tuning amount is determined as it is based on the response waveform having a small amplitude as shown in FIG. 7 (b) and is easily affected by noise, it may give an unstable control characteristic. There is.

The present invention has been made in view of the above circumstances, and an object thereof is to perform self-tuning adjustment capable of tuning an optimum calculation parameter without being affected by noise even when the amplitude of a response waveform becomes small. To provide the total.

(Means for Solving Problems) FIG. 1 is a basic functional block diagram of the device of the present invention. In the figure, reference numeral 1 is a block showing a control target (process), and its dynamic characteristics are assumed to change due to changes in production amount, changes in control target values, various disturbances, and the like. 2 is at least P, I calculation is performed on the deviation signal DV between the process amount PV from the controlled object 1 and the control target value SV, and the operation amount MV obtained
Control means for outputting the output to the controlled object 1, 3 is the process amount PV
Alternatively, the parameter tuning means 4 for tuning the PI calculation parameter of the PI control means in accordance with the response waveform of the deviation signal DV is a tuning amount correction means for correcting the tuning amount of the calculation parameter in relation to the size of the response waveform.

(Operation) The tuning amount correcting means 4 detects the magnitude of the response waveform (for example, the maximum peak value), and corrects the tuning amount of the P, I calculation parameter so that the closer to a good control state, the smaller the tuning amount. . As a result, the PI control means
The optimum calculation parameters are always tuned, and good control can be performed.

(Embodiment) FIG. 2 is a functional block diagram of an embodiment of the apparatus according to the present invention. The same parts as those in FIG. 1 are designated by the same reference numerals. In the parameter tuning means 3, 31
Is a waveform observing means for observing the waveform of the process signal or the deviation signal (here, the deviation signal DV) and outputting some evaluation indexes representative of the response waveform. 32 is a target value (optimal response model) which is a target of controllability. ) Target setting means, 33
Is a parameter calculation means for calculating a PI calculation parameter such that the evaluation index of the response waveform from the waveform observation means 31 becomes a response target preset in the target setting means 32. Based on the calculation result, PI control is performed. The PI calculation parameter of the means 2 is changed / set.

In the tuning amount correcting means 4, 41 is an amplitude value detecting means for detecting the magnitude (for example, the maximum peak value) of the process amount or the deviation signal (here, the deviation signal DV), and 42 is the magnitude x of the deviation signal DV. A correction coefficient calculation means for calculating the correction coefficient α by 43, and 43 is a correction coefficient output means for outputting the correction coefficient α to the parameter calculation means 33 and setting it in the PI control means 2.
The calculation parameter is corrected (corrected) according to the correction coefficient α.

FIG. 3 is a block diagram of hardware for realizing the functions shown in FIG. In the figure, 5 is a controller, which inputs the process amount PV and the control target value SV from the controlled object and outputs the operation signal MV to the controlled object 1. In the controller 5, 51 is a multiplexer that sequentially selects and inputs the process amount PV, various analog signals ei, and control target value SV, 52 is a comparator that receives the signal selected by the multiplexer 51 as one input, and 53 is a micro. On the processor
A signal or the like from the comparator 52 is input. Reference numeral 54 is a D / A converter that converts a digital signal from the processor 53 into an analog signal, and 55 is a sample hold circuit that holds the output of the D / A converter 54 at a predetermined timing. It becomes the operation signal MV. 56 is an I / O port, 57 is a RAM storing various data, 58 is a microprocessor 53
System RA that stores programs for major operations performed by
M and 59 are, for example, ROMs that store programs created by the user according to processes, and 60 is a display keyboard, which are coupled to the microprocessor 53 via the data bus BS.

Here, the system ROM 53 contains the microprocessor 53
However, the waveform observing means 31 and the parameter calculating means 3 shown in FIG.
3, a tuning amount correcting means 4, a program for functioning as the PI control means 2, and various data for functioning as a target setting means 32 including data of an optimum response model of the controlled object 1 are stored. System ROM58, RO
M59 may be shared by one ROM.

The main operation of the device thus configured will be described below.

FIG. 4 is a waveform diagram showing changes in the deviation signal DV when the controlled object 1 changes.

Waveform observation means 31 in the parameter tuning means 3 and amplitude value detection means 41 in the tuning amount correction means 4
For example, a deviation signal DV as shown in FIG. 4 is input, and an evaluation index representative of the waveform pattern and its amplitude value (peak value x) are obtained.

Here, the waveform observing means 31 obtains an area ratio AR calculated as follows, for example, as an evaluation index represented by a waveform pattern. That is, the area A1 of the deviation signal DV to generation time of the second peak lasts from time point of generation of the first peak DV1 of the deviation signal _{DV (t 1) (DV2)} (t 2), generation time of the second peak t _The area A2 related to the deviation signal DV from ₂ to the subsequent time t ₂ + (t ₂ −t ₁ ) is calculated (see FIG. 4). Then, the area ratio AR is calculated from the ratio of the areas A1 and A2 obtained by the calculation.

An example of formulas for calculating the areas A1, A2 and the area ratio AR (1),
This is shown in equations (2) and (3).

In addition, the above-mentioned area ratio represented by a waveform pattern
Other than AR, maximum overshoot (overshoot) OVS, area
The time TA required to obtain A1 and A2 is expressed by equation (4), (5)
Calculate according to the formula.

TA = 2 · (t ₂ −t ₁ ) (5) In this way, the waveform observing means 31 has areas A 1 and A 2 related to the deviation signal waveform that continues from the time when the first and second peaks of the deviation signal occur.
If the area ratio AR using is used as one of the evaluation indices of the waveform pattern, it is not necessary to detect the third peak, so that there is an effect that it is not affected by noise (the third ratio).
The peak is generally smaller in amplitude than the first and second peaks, so it is difficult to distinguish it from noise).

Here, the arithmetic expression of the equation (1) is for obtaining the shaded area A1 in FIG. 4, and the arithmetic expression of the equation (2) is the shaded area A2 in FIG. It is what you want. Therefore, for example, the area A1 is related to the first peak (DV1) of the response waveform and the time (t ₂ −t ₁ ) from the appearance of the first peak to the appearance of the second peak. There is. Similarly, the area A2 is the second peak (DV2),
It is related to the time (t ₂ −t ₁ ) from the appearance of the first peak to the appearance of the second peak.

The calculation by the formula (3) is for obtaining the ratio of the areas A1 and A2, and the area ratio AR obtained by this calculation formula
From (= A2 / A1), the attenuation amount (damping amount), which is well known as one of the evaluation indexes of the controlled object (process)
Information is obtained that is equivalent to.

The calculation by the equation (4) is performed by using the first peak and the second peak of the response waveform.
This is for obtaining the ratio with the peak of, and information about the amount of overshoot, which is well known as one of the evaluation indices of the controlled object, can be obtained.

Further, the calculation by the equation (5) is for obtaining the cycle of the response waveform, and this obtains information about the vibration cycle known as one of the evaluation indexes of the controlled object.

In this way, if the area ratio AR calculated using the above-mentioned areas A1 and A2 of the response waveform is used to obtain information corresponding to the damping amount of the controlled object, in the response waveform, For example, even if the noise NS is mixed in as shown by the broken line in FIG. 4, the area A2 of the response waveform is such that the area where the amplitude is increased and the area where the amplitude is decreased due to superposition of noise cancel each other out. Therefore, it is not affected by noise mixing.

Parameter calculating means 33, each evaluation index of the waveform pattern given from the waveform observing means 31, AR, OVS, TA, and each data of the optimal response model stored in advance in the target setting means 32, Calculate PI calculation parameters.

On the other hand, in the tuning amount correcting means 4, the correction coefficient calculating means 42 inputs the amplitude value x detected by the amplitude value detecting means 41 and calculates the following correction coefficient α according to the amplitude value x.

x is a given dead band (threshold) DB x ≧ 2 · DB
When α = 1 DB>x> 2 DB α = K · x FIG. 5 is a diagram showing the relationship between the amplitude value x and the correction coefficient α. The correction coefficient α is a value smaller than 1 which is proportional to x within the range where the amplitude value x is DB>x> 2DB, and x is x.
It becomes 1 when ≧ 2 · DB.

The correction coefficient output means 43 outputs the correction coefficient α calculated in this way to the parameter calculation means 33.

The parameter calculation means 33 receives the correction coefficient α and calculates the calculation parameters (proportional calculation parameter PB, integral calculation parameter TI) to be set in the PI control means 2 according to the following equation, for example.

PB2 = (1 + α * EAR) * PB1 (6) TI2 = (1 + α * EOVS) * TI1 (7) However, PB2 is the proportional calculation parameter that is set next time PB1 is the proportional calculation parameter that is set this time TI2 is the integration that is set next time The calculation parameter TI1 is the integration calculation parameter that is set this time. EAR is the error area ratio. EOVS is the error overshoot. Α is the correction coefficient. Here, the correction coefficient α is the maximum amplitude x of the response waveform.
The relationship as shown in FIG. 5 is obtained from the experience so far and the results of experiments on various controlled objects. That is, the amplitude of the response waveform is
When it is relatively small, such as when it is between
Since it is easy to be influenced by the mixed noise, the value of the correction coefficient α is set to a value smaller than 1, so that the tuning amount is made smaller (closer) than usual.

Also, receiving such a correction coefficient α smaller than 1,
The calculations of the expressions (6) and (7) performed by the parameter calculation means 33 have the following meanings. In addition,
In the equations (6) and (7), * indicates a multiplication code.

That is, the PI control means 2 is preset with a certain proportional calculation parameter or integral calculation parameter. The response waveform observed by the waveform observing means 31 is observed as the control result by the operation signal (control output) MV calculated based on the proportional operation parameter PB1 and the integral operation parameter TI1 set in the PI control means 2 this time. It If the response status of the observed waveform is the same as the desired response waveform set in the target setting means 32, the proportional calculation parameter PB1 and the integral calculation parameter TI1 currently set in the PI control means 2 are changed. You don't have to. However, when the desired response waveform is not obtained, the proportional calculation parameter PB2 and integral calculation parameter TI2 set in the PI control means 2 next time are
The calculation is performed using the equations (6) and (7).

In the equation (6), the error area ratio EAR is the area ratio of the desired response waveform set in the target setting means 32 and the area ratio AR ((3 of the response waveform actually observed by the waveform observing means 31. ), The error overshoot EOVS is
It is an error between the desired response waveform overshoot set in the target setting means 32 and the response waveform overshoot OVS (calculated by the equation (4)) actually observed by the waveform observing means 31. .

Therefore, in the equation (6), for example, when the vibration of the response waveform is large and the area ratio AR is larger than the desired response waveform (the vibration of the response waveform is large), the next proportional calculation parameter PB2 becomes large (thus, The proportional gain becomes smaller), and the calculation parameter is updated to suppress the vibration of the response waveform.

Similarly, in the equation (7), for example, when the overshoot of the response waveform is larger than the desired overshoot, the next integral calculation parameter TI2 becomes large and is updated to the calculation parameter in the direction in which the integral operation is ineffective. The Rukoto.

Then, in each of these arithmetic expressions, when the correction coefficient α is given, these correction coefficients are respectively corrected so as to be smaller than the normal tuning amount (the state of α = 1).

The PI control unit 2 performs PI calculation using the PI calculation parameter set and changed by the parameter calculation unit 33, and outputs the calculation result to the control target 1 as the operation signal MV.

In the present invention, the calculation parameters PB2 and TI2 set and changed in the PI control means 2 are calculated according to the equations (6) and (7). Therefore, even if the evaluation index of the response waveform does not change, the deviation When the amplitude of the signal DV is small, the correction coefficient α is also small, and as a result, the amount of change in the calculation parameter is corrected so as to be smaller than in the normal case, thereby preventing unnecessary tuning.

In the above-mentioned embodiment, the waveform observing means obtains the area ratio AR as one of the evaluation indicators of the response waveform, but it is also possible to obtain the damping ratio calculated from the peak value. Further, the waveform observing means 31 and the amplitude value detecting means 41 are both adapted to input the deviation signal DV, but if the control target value SV is constant, the process amount PV is input. You may Further, the calculation formula of the correction coefficient α and the calculation formula of the parameter calculation means 33 are not limited to those described above. Further, in the above description, the controller having the PI control means is described as an example, but it may be applied to the controller having the PID control means. The correction coefficient α is changed according to the amplitude x of the response waveform here, but may be changed according to other conditions.

(Effects of the Invention) As described in detail above, according to the present invention, even when the amplitude of the response waveform is relatively small, it is not affected by the noise mixed in the response waveform and is always optimal. The calculation parameters can be tuned.

[Brief description of drawings]

FIG. 1 is a basic functional block diagram of the device of the present invention, FIG.
FIG. 3 is a functional block diagram of an embodiment of the apparatus according to the present invention, FIG. 3 is a block diagram of hardware for realizing the functions shown in FIG. 2, and FIG. 4 is a deviation signal when the controlled object 1 changes. FIG. 5 is a waveform diagram showing a change in the above, FIG. 5 is a diagram showing the relationship between the amplitude value x and the correction coefficient α in the correction coefficient calculating means
FIG. 7 is a diagram for explaining the operation of the conventional device, and FIG. 7 is a diagram for explaining the problems in the conventional technique. 1 ... Control object, 2 ... PI control means, 3 ... Parameter tuning means, 4 ... Tuning amount correction means.

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Takashi Kono 2-932 Nakamachi, Musashino-shi, Tokyo Inside Yokogawa Electric Co., Ltd. (56) Reference JP-A-61-245203 (JP, A) JP-A-SHO 60-176104 (JP, A) US Patent 4602326 (US, A)

Claims (1)

[Claims] 1. A parameter tuning means for observing a response waveform of a process amount or a deviation signal between a process amount and a control target value, and adjusting a calculation parameter to an optimum value based on an evaluation index of the response waveform. In a controller equipped with: a tuning amount correction means for calculating a correction coefficient α for correcting the tuning amount of the calculation parameter to be smaller than the normal tuning amount when the magnitude of the response waveform becomes smaller than a predetermined value. the provided, said parameter tuning means, the response waveform (DV) first third of second peak lasts from time of occurrence of the peak point of (DV ₁₎ (t ₁₎ of the (DV ₂₎ the time of occurrence (t ₂₎
The area of the response waveform to (A1), obtains the area (A2) according to said subsequent from the generation time point _{(t 2) t 2 + (} t 2 -t 1) response waveform to time was determined area (A1), (A2)
The self-tuning controller is characterized in that the area ratio (AR) is calculated and the area ratio (AR) is multiplied by the correction coefficient (α) obtained by the tuning amount correcting means to tune the calculation parameter. .