JPS6015705A - Automatic control method of control parameter - Google Patents

Abstract

PURPOSE:To decide an optimum control parameter with the retrieval of just one time to obtain the control characteristics which excel in the stability and the responsiveness by keeping the operation quantity within a limited range and preventing the overload of a device such as a pump, etc. of an operation part. CONSTITUTION:A process identifying part 7 identifies the coefficient of a self-regression shift average model by a Kalman filter algorithm and based on the time series data on the operation quantity v(t) and control quantity y(t), i.e., the input and the output of a process 1. Thus a transmission function Gp(s) of the process 1 to be controlled is obtained. A reference model deciding part 10 supplies the identified function Gp(s), a reference model Gm0(s, sigma0) related to the quantity y(t) and a reference model Gm1(s, sigma1) related to the quantity v(t) to decide the rise time sigma0 of the y(t) and to deliver a reference model Gm0(s) related to the y(t). A parameter deciding part 11 decides control parameters Kp and Ti so that the coincidence is obtained between said model Gm0(s) and a transmission function w(s) of a closed loop containing the target value yr(t) through the quantity y(t).

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a method for optimally adjusting control parameters of a process control device, and in particular, to maintain stability and coordination by maintaining the manipulated variable within a limited range. The present invention relates to a control parameter automatic adjustment method suitable for obtaining control characteristics aimed at.

[Background of the invention]

FIG. 1 shows a conventional control parameter automatic adjustment method. 1
is the process to be controlled, 2 is a computing unit that computes the deviation between the controlled variable y(t) and its target value y, and 3 is a control device that performs control calculations such as PID calculations. t is time , S
is the Laplace operator.

Reference numeral 9 denotes an optimum control parameter determination unit; 7 a process identification unit that identifies the transfer function a p (S) of the process 1; and 8 determines the optimum value of the control pager in the transfer function G of the control device 3 (s). This is a parameter determination section.

The control device 3 inputs the target value y, (t) and the control amount y (deviation of ri (y, (t) - y(t)) and outputs the manipulated variable v(t) for the process 1. However, if the control device 3 performs PI calculation, its transfer function G,
(S) is the following formula % formula % (1) Here, is a proportional gain and TI is a control parameter called integral time. In order to quickly match the control amount y(ri) to the target value y・(ri) and obtain stable control characteristics,
It is necessary to adjust the above control parameters to optimal values according to the process l.

Therefore, the process identification unit 7 identifies the transfer function op(S) of the controlled process l by inputting the manipulated variable v(t), which is the input/output of the process, and the control my<, and the parameter determination unit 8 , the control parameters and the optimum value of TI are determined based on the above a p (S). These optimal control parameters are input to the control device 3, and the values of the control parameters are corrected to the optimal values.

The optimal parameter determining method in the parameter determining section 8 will be described below. This method is called the (partial) model matching method, and is described in ``Control system design method based on partial knowledge of the controlled object'' (Proceedings of the Society of Instrument and Control Engineers, Vol. 15, No. 4).
Since it is described in detail in 1982-8), an outline will be given here. In the model matching method, the ideal transient characteristic Gmo( This is a method of determining the parameters of the transfer function G, (s) of the control device 3 so as to match the value S+σ.).

W(S) and G-force 0(S, σ0) are as follows.

・・・・・・・・・(3) Here, Y(a), y, (s) are respectively il,
y(*y r(t)) is subjected to Gibras transformation. σO is the amount corresponding to the start-up time of the control amount y(ri). α2゜α3.α4 is a constant, (respond quickly without 2-bashoot) c to j
-L, α2=0.375. αg=0.0625. α4=
Set it to 0.0039. Transfer function G of controlled process l
, If (S) is as shown in equation (4), then equation (5) holds true if equation (3) matches.

When the denominator of equation (5) is divided by the numerator, the following equation holds true.

+(α22-α3)σo")B 10...] ・・・・
...(6) On the other hand, the molecule of G and (S) is a linear equation as shown in equation (1).

Therefore, equations (1) and (6) are matched as far as the adjustable angle meter allows. Therefore, the following formula holds.

go g2 gx 0= (−1α2σo−+(α22−α3)σo”)・
...(9) σo go g.

According to equation (9), find σo (minimum positive value that satisfies equation (9)) for the start-up time of the controlled variable, substitute σ0 into equation (8), find, and convert these into (7 ) to calculate TI. According to this method, the closed loop transfer function W(s) matches up to the third-order term of the reference model 0□ in equation (3) and S in (8). In this way, as long as the low-order terms of S match, the control amount y(i) exhibits ideal control characteristics.

Therefore, in the parameter determination unit 8, conceptually, the identified a p <8) and the reference model Gmo(S+σ.)
By inputting , , TI is determined as the parameter of G,(s). Specifically, as shown in Figure 2 (
Find the minimum positive σ0 in equations (9) and Kp+T+ in equations (8) and (7).

FIG. 3 shows the control parameter Kp using the method described above.

The step response characteristics when adjusting TI are shown.

However, the transfer function of the controlled process was set to the following formula.

1 °, (8) = ...Mountain (1o) 1,000 4s + 2.
4s2 When the target value Y, (t) is increased stepwise from O to 1, the control amount y (represents ideal characteristics in which the reference model Gmo (Ssσ0) and t"t match. However,
The manipulated variable v(t) greatly exceeds the rated value (1,0). For example, if the operating part is a pump, this means that the rotation speed of the pump will exceed the rated value. There is. Therefore, in reality, the characteristics shown in FIG. 3 cannot be achieved in many cases because the manipulated variable V(R) is greater than the rated value.

8. In other words, the conventional automatic control parameter adjustment method using the model matching method described above has the advantage of being able to determine control parameters in a single search (in the past, they were often determined by trial and error). However, there is a drawback that it is impractical to have the manipulated variable exceed the rated value, and in other words, when the operating section is a device such as a pump, the second problem is that the device will be overloaded.

[Purpose of the invention]

The purpose of the present invention is to set control parameters to obtain control characteristics with excellent stability and responsiveness under the condition that the manipulated variable V is maintained within a limited range (for example, within 1.2 times the rated value). An object of the present invention is to provide an automatic adjustment method.

[Summary of the invention]

In the present invention, in the conventional model matching method, the control amount Y(t
), paying attention to the fact that the manipulated variable v(t) becomes greater than the limit value if the startup time σ0 of ) to the manipulated variable v(t) is determined so that the transfer function V(8)/Y, (s) matches the reference model regarding the manipulated variable, and the reference model regarding the controlled variable determined based on the above σoK is determined. The parameters of the control device are determined so that the closed loop transfer function W (S) from the target value y r (t) to the controlled amount y(t) matches. Therefore, the manipulated variable v(t) exhibits characteristics that match the characteristics of the reference model regarding the manipulated variable, and the control 1y(t) exhibits characteristics that match the characteristics of the reference model regarding the controlled variable.

[Embodiments of the invention]

FIG. 4 shows an embodiment of the present invention. 20 is an optimal parameter determination unit, 7 is a process identification unit, 10 is a transfer function op (S) of the identified controlled process, a reference model Gmo (S + σ0) related to control jty (), and operation Mv (
A reference model that inputs the reference model G++++(S+σl) related to t), determines the short rise time σ0 of the controlled variable Y(t), and outputs the reference model a ta o (s) related to the control variable Y(t). This is the decision making section. Here, σl is an amount corresponding to the short time required for the rise of the manipulated variable v(t). 11 is the target value y, and the closed loop transfer function W(8) from the control amount y(ri) to the control amount y(ri) is the reference model G-o for the control amount y(ri).
This is a parameter determining unit that determines the control parameters of the control device 30 so as to match (5).

The process identification unit 7 uses the Kalman filter algorithm to identify the coefficients of the autoregressive moving average model based on the time series data of the manipulated variable V (and the controlled variable y(t), which is the input/output of the process 1). Transfer function a p of process 1
Enter (S).

The reference model determining unit 10 determines the transfer function V(S)/Y from the controlled variable y (the short rise time σ0 of the R2, target value y) to the manipulated variable v(t) as follows. (S)
is expressed as equation (11). However, V[S) is v(t)
is the Laplace transform of

The closed loop transfer function W(S) = Y(S)/Y, (S) from the target value yt(ri to the controlled variable y(t) is the reference model Gyo(8, σ0) for the controlled variable y<t>. ), equation (11) becomes equation (12).

・・・・・・・・・・・・(12) When the denominator of equation (12) is divided by the numerator, equation (12) becomes (13)
It becomes like the expression.

・・・・・・・・・(13) Reference model G11ll(Slσ1
) is defined as follows.

g.

G...(゛・°・)=□10,18+βz('ts)・
-°--- (t4) Here, β2 is a constant, and when the response characteristic of the manipulated variable v(t) is made to be the same as the response characteristic of the controlled variable y(t), it is set as β2−α2. (13) and (
14) Make the primary and secondary coefficients of S equal to each other so that the equations match as closely as possible. Therefore, the following formula holds.

Equation (15) and the following equation hold true.

(16) Find the minimum positive σ0 that satisfies equation (16). Once σ0 is determined, the reference model G-o(S) regarding the control amount y(t)
is determined as follows.

G-o(s)=1/(1+σQS+α2σo2.ff+
α3σ, 3S3+...) (17) (11) Therefore, the reference model determining unit 10 specifically determines the minimum positive σ0 that satisfies equation (11) as shown in FIG.
Perform calculations to calculate the difference.

Next, the parameter determination unit 11 determines the closed loop transfer function W
(8) is the reference model G, Q (6) shown in equation (17)
The parameters of G and (8) are determined so that they match. When the control device 3 is a PI@ calculation, G and (S) are as shown in equations (1) and (6), so equations (7) and (8) hold true. Therefore, for the parameters of the PI calculation, 'l
”、Ha(7n style.

It can be determined from equation (8).

In this case, the closed-loop transfer function W (S) is the reference model a
rm o (s) and the second-order term of S match.

Therefore, as shown in FIG. 6, the parameter determination unit 11 corrects the parameters Kp and T of the control device 3 using s'rl according to the calculation formula (18).

FIG. 7 shows step response characteristics according to an embodiment of the present invention. The controlled process is the same (10) as in Figure 3.
The formula was β2=α2=OJ75.

The manipulated variable v (t) is the rated value v (t) is the rated value (1,0)
The start-up time σ0 of the controlled variable y is longer than that in the conventional case shown in Fig. It is possible to obtain control characteristics of the control amount y (y) with excellent responsiveness.

According to the embodiment of the present invention described above, operation 1v(t)
It is easy to find the optimal control parameters to obtain control characteristics with excellent stability and tolerance while maintaining the control parameters within the limit range and preventing overload of equipment such as the pump e in the operating section. There are two benefits to being able to decide.

FIG. 8 is a diagram showing a modification of the present invention, and the difference from FIG. 4 is that a noise filter 4 is added,
The difference lies in the content of the calculation in step 3.

- If the transfer function of the noise filter 4 is G t (s), the closed loop transfer function W(8) is as follows.

If the noise filter 4 is a first-order lag, G t (S) will be as follows. Ts is the time constant of the filter.

Gr(s)= □ ・・・・・・・・・(20)1+T
fs Reference model G of equations (19) and (17). Assuming that (S) matches, equation (19) becomes as follows.

) By substituting equations (4), (17), and (20) into equation (21), G and (8) become as follows.

〉 0 ・・・・・・・・・(22) When the control device 3 performs PI calculation, (Divide the denominator of equation 221 by the numerator D, make it match equation (1) K), T
+ becomes as follows.

Therefore, in FIG. 8, the reference model determining unit 10 performs the calculations shown in FIG. 5, as in the case of FIG. 4. The parameter determining section corrects the control parameters of the control device 3 using equation (23), including 9Kp and T+, as shown in FIG. According to this modification, even if something like the noise filter 4 is added, the optimum parameters can be easily determined.

FIG. 10 is a diagram showing another embodiment of the present invention.

The control system consists of an I-P that includes an integrator 5 that performs integral action, a feedback element 6 that performs proportional action and differential action, etc.
This is the D control system. The process identification section 7 and the reference model determination section 10 perform the same calculations as in the case shown in FIG. The parameter determination unit 15 determines optimal control parameters based on the following principle. The closed loop transfer function W(s) is expressed as equation (24).

Here, the transfer function of the foot coupling element 6 is defined as F (S
). Reference model G of equations (24) and (17)
If −o(8) matches, then F(8) becomes as follows.

= (kα1σo go) 10 (kα2σo2gt) 80 (
kα3σo”gz)82+・・・・・・ ・・・・・・
(25) If F (S) has only proportional action, then F'(S
) becomes as follows. fo is a proportional gain.

F(s)=fo ・・・・・・・・・・・・(26)(
When equations (25) and (26) are made to match as much as possible, the following equation holds true.

fo = α1σo go ・・・・・・・・・・・・(
27) 0 = α2σo” gx ・・・・・・・・・・・・
...(28) Substitute the formula (28) and convert this into (27)
Substituting into the formula will yield f. Therefore, the calculation contents of the meter determining section 15 are as shown in FIG.

According to another embodiment of the present invention, in an I-PD control system, a control characteristic with excellent stability and responsiveness can be obtained under the condition that the manipulated variable v(t) is maintained within a limited range. This has the effect of automatically adjusting optimal parameters.

〔Effect of the invention〕

According to the present invention, it is possible to easily determine and automatically correct optimal control parameters for obtaining control characteristics with excellent stability and responsiveness within the condition that the manipulated variable is maintained within a limited range. .

[Brief explanation of the drawing]

Figure 1 is a diagram explaining the conventional control parameter automatic adjustment method, Figure 2 is a diagram supplementary to Figure 1, Figure 3 is a diagram explaining the control characteristics of the conventional method, and Figure 4 is a diagram explaining the control characteristics of the conventional method. FIG. 5 and FIG. 6 are diagrams supplementing FIG. 4. FIG. 7 is a diagram explaining control characteristics according to an embodiment of the present invention. FIG. 8 9 is a diagram explaining the present invention in detail, FIG. 9 is a diagram supplementary to FIG. 8, and FIG.
0 is a diagram for explaining another embodiment of the present invention, and FIG. 11 is a diagram for supplementary explanation of FIG. 10. 1... Controlled object, 3... Control device, y-(')...
・Target value, y(su...Controlled amount, W(8)...Closed loop transfer function, G-o(s). G+++o(S+σ0)...Reference model regarding controlled amount, Gmx(S, σl) ...Reference model regarding the amount of operation, σ0 1 [i: 1 2 necessary 3 l B saki mon (rare 5 s 50 50 tsu 0 (2T 7 gu 88 questions αri 80)

Claims (1)

[Claims] 1. In a method of identifying a transfer function of a controlled object and adjusting control parameters of a control device, a closed-loop transfer function from a target value to a controlled variable matches a first reference model regarding the controlled variable; The control parameter automatic adjustment method is further characterized in that the control parameter of the control device is adjusted such that a transfer function from the target value to the manipulated variable matches a second reference model regarding the manipulated variable. 2. Determine the rise time σ0 of the first reference model (3m o (S)) to match the second reference model,
The control parameter automatic adjustment method according to claim 1, wherein the control parameters of the control device are adjusted so that the closed loop transfer function from the target value to the controlled variable matches the first reference model G-o(s). .