KR950002690B1 - Automatic tuning controller and method of determining pid parameters - Google Patents

Abstract

No content.

Description

Automatic Tuning Controller and Determination of Proportional, Integral, and Derivative (PID) Parameters

1 is a block diagram showing one embodiment of the present invention.

2 is the limit that occurs during tuning. Drawing showing the cycle.

3 is a block diagram showing a conventional example.

4 is a view showing a limit cycle resulting from the configuration of FIG.

5 is a diagram showing an occurrence point of a limit cycle.

6 is a block diagram showing an embodiment in which the controller is composed of a microcomputer.

7 and 8 show an example of a processing program in the CPU.

* Explanation of symbols for main parts of the drawings

1: processor 10: controller

11: proportional calculation unit 12: integration operation unit

13: differential operation unit 14: output unit

15 input unit 16 observer

17: state transition control unit 18: nonlinear requirements

19: regulator SW1-SW3: switch

The present invention performs a proportional integral derivative (PID) operation on the deviation e between the control amount pv and the set value sp fed back from the control object (process) in the feedback control system, and processes the obtained operation amount m in the process. The present invention relates to a method for determining PID parameters for achieving optimum control of a controller and a control object using the same.

PID controllers are widely used in the industry, but the adjustment of PID parameters, which is the premise of the optimum control thereof, is generally performed manually, and step response and limit sensitivity methods are well known. However, when the characteristics of the process change during operation, it is difficult to quickly correct the optimum parameters, which takes a long time to measure the characteristics, and the control is stopped while the characteristics are measured. (pv) does not become a preferable value.

In addition, there is a device for automatic adjustment of PID controller, but this device is expensive and not easy to use.

In addition, there is an automatic PID controller, but such a device is more complicated than a normal PID controller, and in reality, it is common to use a computer, but the operation cycle (sampling / time) is long, or program memory is used. Will cause an increase in capacity.

On the other hand, a method has been proposed for a long time to allow the controller to have a nonlinear characteristic and cause a limit cycle in the process. They show the configuration of the controller in the case of tuning (acoustic, wavelength adjustment) as shown in FIG. 3, and the nonlinear element 2 is discontinuous control operation (2 position control for deviation). It is introduced to the signal so that it is limited. In FIG. 4 and when the guy limit and the cycle occurred, the characteristics and the optimum parameters of the process 1 could be easily obtained from the wavelength thereof. Furthermore, in the figure, (a) means manipulated amount and (b) means deviation.

However, in this case, if the position value M of the nonlinear element 1 is too large, the variation of the manipulated variable becomes excessively large, except that a rapid response such as some temperature control systems is not required. , Shows a downward shape.

On the other hand, the limit cycle uses the processor's equilibrium point (m when sp = pv) as an operation starting point. There was a case where a cycle did not occur. In addition, pv at that time is not desirable to stay at a place far from s.

As a result, such a conventional method lacked practicality.

The present invention provides a non-linear element, at the same time inserts the non-linear element into the front end of the proportional calculation section, and provides switch means for transitioning to the tuning state in which the integral calculation unit is connected in parallel with them, and the limit occurring under the tuning state again. The observer of the cycle, the regulator which determines the optimum PID parameter from the observation result and sends it to each operation unit, operates the switch means by establishing predetermined tuning initiation conditions in normal operation state, and at the same time determines the PID parameters under the tuning state. The control means for restoring the means of the switch is provided.

In addition, the present invention generates a limit cycle while continuing the feedback control of the process while inserting the nonlinear elements at the front end of the proportional calculation section and connecting them to the integral calculation units in parallel, thereby generating a limit cycle and the observation result thereof. To get the PID parameters.

From the observed point of the limit cycle, the desired limit cycle can be calculated, from which the characteristics of the process, and hence the optimum PID parameters, are obtained.

Normal PID control is performed during normal operation, but when the deviation exceeds a predetermined value, for example, the PID controller automatically shifts to the tuning phase, and in order to achieve optimum control of the process, the PID parameters are corrected and then updated by the updated parameters. Return to normal operation. During tuning, feedback control of the process is maintained.

1 is a block diagram showing one embodiment of the present invention. In this figure, the automatic tuning controller 10 of this embodiment includes three switches SW1, SW2, and SW3. These switches are linked to each other and are tuned when pressed in the A width (Auto) and in the normal operation state, and pushed in the T side (Tuning).

The controller 10 outputs a proportional calculation unit 11, an integral calculation unit 12, a differential calculation unit 13, and an operation amount m obtained by adding the outputs of these calculation units to the processor 1. And an input unit 15 which obtains the deviation e between the control amount pv fed back to the process 11 and the set value sp and supplies it to each of the calculation units, in order to eliminate the deviation e in the normal operation state. The manipulated variable m obtained by expression (1) is output.

m = m ₁ + m ₂ + m ₃

………………………………………(1)

… … … … … … … … … … … … … … … (One)

K, Ti, and Td are proportional gain, integral time, and derivative time, respectively, and they are called PID parameters. M is the output ₃ of the right side first term proportional operation section 11 outputs m _1, the second term output m _2, paragraph 3 of the differential computing section 13 of the integration operation section 12 of the. Depending on the nature of the process and the purpose of the control, various modifications are possible, such as eliminating the derivative term (PI control) or acting proportionally and the derivative term only on pv (PID control). It is called controool.

In this way, the controller 10 performs a general PID controller during normal operation, but at that time, the observer 16 monitors the movement of the deviation e. Thus, the absolute value of the deviation monitored by this observer 16 becomes larger than a predetermined value. Or when the predetermined tuning start condition is satisfied, such as the responsiveness becomes unsatisfactory, the state conversion control part 17 and switch SW1-SW3 are switched to T side, and the controller 10 moves to a tuning state. .

In the tuning state, a nonlinear element 18 having a nonlinear characteristic at two positions is inserted at the front end of the proportional calculation section 11, and the integral calculation section 12 is connected in parallel with them. That is, K and Ti are both, and Td = 0, and the differential operation is turned off, and the operation amount m is the sum of the outputs according to the following expressions (2) and (3).

At this time, there is no change in operating so as to eliminate the deviation e. That is, the feedback control system of the process is still maintained.

FIG. ₂ shows wavelengths of the manipulated variable m and the deviation e at this time, and the output power m ₁ and m ₂ of the positive calculation unit. In FIG. 2 (a), (a) shows m and (b) shows e, and in FIG. 2 (b), (a) shows m ₁ and (b) shows m ₂ . The observer 16 monitors their movement.

That is, the vibration of the m ₂ sufficiently jakeunga, the vibration of the or monitoring the well jakeunga compared to m, when the both of these conditions does not satisfactory and less K via an adjuster (19), increasing the Ti in comparison to m ₁ . This is for safe preservation of the process state when K and Ti are abnormal. In particular, when starting operation, it is sometimes impossible to know how much of the parameter is appropriate. In such a case, the operation is important.

If the monitored operation satisfies each of the above conditions, the vibration of the deviation e is kept at a stable continuous vibration, that is, a limit cycle, and then the amplitude X of the vibration and the period Tc are observed, and the data is adjusted by the regulator 19. Send it out.

In the regulator 19, the conventional limit sensitivity Kc and the vibration period Tco 'at that time are simply obtained by the following calculation.

Where M is the value of the two positions of the non-linear element 18.

Thus, the fact that Kc and Tco 'are represented by said Formula is demonstrated next.

의점이, 비선형요소(18)와 비례연산부(11)만에 의한 제거계에 있어서의 리미트.사이클(이를 바람직한 리미트 사이클이라고함)의 발생점이다. 이 바람직한 리미트.사이클의 진폭 X₀와 주기 Tco가 얻어지면 Kc 및 Tco'는 다음식에서 얻어진다.5 is a diagram showing a limit cycle occurrence point. In FIG. 5, the transfer coefficients Gp (jw) of the process and (b) record the controller 10 when Ti is set to an appropriate value. Nyquist plot of -1 / N from the coefficient N (X, w). (A) and my intersection marked with ○ in figure are limit. It is point of occurrence of cycle, but phase -180 of Gp displayed with ● in figure in same figure

The point of interest is the point of occurrence of a limit cycle (referred to as a preferable limit cycle) in the removal system by only the nonlinear element 18 and the proportional calculation unit 11. When the amplitude X ₀ and the period Tco of this preferred limit cycle are obtained, Kc and Tco 'are obtained from the following equation.

Kc = 4KM / πX ₀ ‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥ 7

Tco ′ = Tco ‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥ 8

However, as mentioned above, it is practically difficult to obtain these data directly from actual processes, that is, continuing process control. Therefore, in the present invention, it is possible to obtain the data of the preferable limit cycle generation point from the data of the limit cycle occurrence point in the tuning state described above.

First, the transfer coefficient Gp (s) (where s is laplace: the laplace operator), which represents the characteristics of the process, is expressed as follows in view of the operation from the equilibrium point.

Gp (S) = e ^-LS / TS ‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥ 9

L: delay time

T: Response Trend

On the other hand, the generated limit cycle is represented by the following equation as a sinusoidal wave.

e = Xsin ωt ‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥‥ (10)

As described above, from the fact that the recording coefficient of the nonlinear element having the nonlinear characteristic in the two positions is 4M /? X, the outputs m ₁ , m ₂ and the manipulated variable m are respectively expressed as follows.

When the vibration component of the control amount pv is e ₀ , the equations (9) and (13) are given.

In this case, puncha e = -e _0, so from (10), (14)

Each frequency ω is converted into a period Tc by ω = 2π / Tc

Next, if the integral operation is also erased in the configuration of the tuning state (that is, Ti = ∞), then the limit cycle at that time is the preferred limit cycle.

e = X ₀ sin ω ₀ t. … … … … … … … … … … … … … … … … … … … … … … (19)

………………………………………………………………(20)

… … … … … … … … … … … … … … … … … … … … … … … … 20

Let period be Tco (= 2π / ω ₀ )

Tco = 4L... … … … … … … … … … … … … … … … … … … … … … … … … (21)

It can be seen that.

Here, instead of finding the L and T of the process itself, we find X ₀ , Tco, from (16), (18) and (20) (21)

의 값이 충분히 작을때)에는 충분히 효과가 있다.As simply X ₀ , Tco is obtained. However, in the process of deriving (22)-(24), tan ^−1θ = θ (θ is an arbitrary angle small enough) is approximated. This approximation, when the system is stable

Is small enough).

Tco를 (7), (8)식에 대입함으로써 전술한 (4)-(6)이 얻어진다.These preferred limits. Data of cycle X ₀

Substituting Tco into the formulas (7) and (8) yields the above-mentioned (4)-(6).

The regulator 19 obtains an optimum PID parameter again from Kc and Tco 'obtained by the formulas (4) and (5), and sends it to each calculation unit. For that purpose, for example, as shown in the following table, the method of "Ziegler.Nichols" is based on already known methods.

In addition, the selection of PI and PID may be performed artificially or automatically depending on the nature of the process as described above.

In addition, if the recording coefficient of the controller 10 in FIG.

……………………………………………………………(25)

… … … … … … … … … … … … … … … … … … … … … … … (25)

Accordingly, FIG. 5 shows the Nyquist curve of -1 / N when Ti is a proper value. However, in general, the amplitude trajectory (i) in the figure along with the increase of X is the frequency trajectory, Limit when penetrating from inside to outside. Cycle is stable.

의 값은 거의 1로 되는 수가 많고, 이 경우에는 (4),In the case where K and Ti are already almost suitable values, i.e., when the stable normal operation is continued, etc.,

The value of is often almost 1, in this case (4),

(5) The calculation speed of equation becomes faster.

After the determination of the PID parameters, the state transition control circuit 17 recovers the switches SW1-SW3 and then performs normal operation using the updated PID parameters.

However, even during tuning, the controller 10 operates so that the puncture e is erased, i.e., the closed loop control system is operated by the feedback, as described above, but the integrating arithmetic unit 12 is connected. Since, for example, even when disturbance occurs in the process, or the SP is greatly changed, and it is impossible to follow only by the nonlinear element 18 when the value of M is small, the integrating calculation unit 12 is effective. Can be erased by changing the starting point control of the vibration caused by the nonlinear element 18. In other words, since the value of M, which is the nonlinear element 18, becomes small, for example, about 1-10% of the transition dynamic range of the manipulated variable is sufficient, there is little variation in the process during tuning, and rapid response Can be secured.

Such a controller 10 can be realized by, for example, executing a program previously stored in a memory by using, for example, a microcomputer. It goes without saying that the proportional, integral and derivative calculation units, the nonlinear elements, the entry and exit units, and the switches SW1-SW3 and the control unit, the observer, the regulator and the like may be realized by individual components having a single function.

From the amplitude X and the period Tc obtained in the observation of the limit cycle generated in this way and the internal constant of the controller 10 at that time, the characteristics of the process and the optimum PID parameters are simply obtained. Returns to normal operation as a PID parameter.

Furthermore, the calculation units of the proportional, integral, and derivative, nonlinear elements, input and output units, normal operation state, tuning state and conversion control means, limit cycle cycle observation means generated in the process during tuning, X and The specific configuration such as a means for obtaining Tc, calculating Kc and Tco ', and calculating new PID parameters from Kc and Tco' is not particularly limited. That is, these may be realized by separate components, or a microcomputer may be used to attempt to realize each of the above functional means as time series execution of a program stored in a memory in advance.

6 shows an example using the microcomputer 20, 21 denotes a processor unit (CPU) such as a microprocessor, 22 denotes a fixed memory (ROM), 23 denotes a variable memory (RAM), and (24). ) Is an input / output port, and 25 is a setting / indicator provided with a setting operator and a display for various constants and the like.

The CPU 21 executes the program stored in the fixed memory 22 in advance in time series to perform the function as the controller 10. This is illustrated in FIGS. 7 and 8.

7 is a flowchart showing an example of a main program processed by the CPU 21. As shown in FIG. In this figure, the CPU 21 enters the set value sp and the process 1 given through the signal path or from the setting / marker 25 in the input processing (step 101) after the initialization processing (step 101). The control amount Pv fed back from is digitally converted and processed through the input / output port 24. Subsequently, the CPU 21 performs a control calculation process to determine the operation amount m (step 103), and in the output processing (step 104), analog-converts the operation amount m through the input / output port 24 to process (1). ) From the above input processing step 102, the output processing step 104 is repeatedly executed at a constant sampling cycle.

8 is a flowchart showing an example of the control operation processing program described above. When the execution of the program shifts to the control operation, the CPU 21 checks (step 201) whether or not the tuning mode is performed by a flag provided by using a predetermined area of the variable memory 23, and the tuning mode. If not, the deviation e between the set value sp and the control amount Pv is determined (step 202), and it is determined whether or not it satisfies the predetermined tuning start condition (step 203). The manipulated variable m is calculated by a normal PID operation using the PID parameters stored in the predetermined area (step 204). That is, this case corresponds to the case where the switch SW1-SW3) is laid on the A side in FIG.

On the other hand, when the tuning start condition is satisfied (step 203), after setting the tuning mode (step 205), the nonlinear element 18 is inserted at the front end of the proportional calculation unit 11 and integrated in parallel with them. Assuming a configuration in which the calculation unit 12 is connected, a limit cycle is obtained (step 206). At this time, the operation amount m is also naturally obtained. Thus, if X and Te of the formula (16) and (18) are obtained from the limit cycle (step 207) (which corresponds to the limit cycle being stable), again, as described above, for example, "Zickler, The optimum PID parameter is determined according to the "Nichools" method, and the value of the PlD parameter up to that point stored in the predetermined area of the variable memory 23 is updated (step 208). Then, the tuning mode is released (step 209). do. Therefore, in the execution of the next PID operation step 204, the operation is performed by the new PID parameter.

In this way, optimal control is maintained while the PID parameters are updated appropriately.

In addition, the case where the deviation e is monitored and automatically shifted to the tuning state by the state thereof has been described. However, other operations, for example, the updating operation of the PID parameters is periodically performed under the tuning start condition as the tuning start condition. It is okay. For example, the operation input from the setting / display section 25 may be added at any time to the state shifting to the tuning state.

As described above, according to the present invention, a nonlinear element, a switch means for inserting the nonlinear element at the front end of the proportional calculation unit and transitioning to the tuning state in which the integral calculation unit is connected in parallel with them, and the limit occurring in the tuning state .The control which operates the said switch means by setting the tuning starter in normal operation state and the regulator which determines the optimum PID parameter from the cycle observer, its observation result, and sends it to each arithmetic unit, and recovers after determination of PID parameter. By setting the means, it is possible to realize a low cost automatic, tuning, and controller that can be used for processes that require simple handling and quick response.

That is, the conventional expensive automatic adjustment device is not necessary at all, and the configuration is simpler than that of the conventional automatic application type controller, and the calculation is simple even when realized by a computer, resulting in an increase in memory capacity. There is no work.

In addition, since the operation amount changes in the process due to tuning, there is little change, and the sampling time is equivalent to that of a normal DDC (dialect.digital controller).

In addition, in use, it is not enough to simply set an appropriate value (a rather large value is good) as the initial value of the PID parameter, to bring it to an appropriate place by manual operation, and then to switch to the normal operation state. If there is no function, it may be started immediately from the normal operation state.

In this regard, the "algorithm" for obtaining X and Tc from the tube axis of the waveform, and the relational equation for connecting Kc and Tco 'with an optimum PID parameter are not particularly limited, and the aforementioned Ziegler-Nicools method is described. Using is just one example of him.

According to the present invention, the optimum PID parameter can be easily obtained from the observation result by inserting a nonlinear element at the front end of the proportional calculation unit and generating a limit cycle while connecting the integral calculation unit in parallel thereto. At the same time, the feedback control of the process can be effectively maintained, and the optimum control can be continued by appropriately updating the PID parameters without breaking control continuity.

Claims (2)

A control unit that performs proportional, integral, and derivative calculations for the deviation between the set value and the control amount fed back from the process, and an output unit for supplying to the process an operation amount obtained by adding the calculation result of each operation unit. In the roller, a switch means for inserting the nonlinear element, the nonlinear element at the front end of the proportional calculation unit, and transitioning to the tuning state in which the integral calculation unit is connected in parallel with the nonlinear element, and the limit occurring in the process in the tuning state. An observer for observing the sine, a regulator for determining the optimum PID parameter to be used for process control from the observation result, and sending it to each operation unit, and operating the switch means by establishing a predetermined tuning start condition in a normal operation state. At the same time recovering the switch means after determination of the PID parameters in the tuning state. Automatic means tuning controller, characterized in that it comprises a control means. In the PID controller that performs PID control on the deviation between the set value and the control amount fed back from the process, and outputs it to the process of the obtained manipulated value, the nonlinear element is inserted at the front end of the proportional calculation unit and paralleled with them. The process control is continued while the integral calculation unit is connected, so that the limit cycles occurring in the process can be observed to obtain the characteristics of the process, and the optimum PID parameters for subsequent process control are determined from the results. A method for determining PID parameters using the apparatus of claim 1.

Images