JPH01224803A - Method for adjusting controlled variable of feedback controller - Google Patents

Abstract

PURPOSE:To simplify an arithmetic processing by sampling-detecting the step response of a process based on the change of a target value substituting a detecting result to the moment approximate expression of dynamic characteristics, enumerating the parameter of the dynamic characteristics and identifying the dynamic characteristics at the time of starting the control and at the time of changing the target value. CONSTITUTION:A step shape signal is given to a controlled system process 4 in a control condition based on the setting and the changing time of a target value U of control starting, and then, the response of the controlled system process 4 is step-changed. Then, the step response of the controlled system process 4 is detected by sampling, a detecting result is substituted to a moment approximate expression, and process dynamic characteristics are identified from a simplified algebraic arithmetic operation. Thus, the dynamic characteristics of the controlled system process is identified accurately and quickly with a simplified processing without executing the complicated processing of a waveform analysis, etc., and a controlled variable can be adjusted automatically.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a control amount adjustment method for a feedback control device equipped with an auto-tuning mechanism.

[Conventional technology]

Conventionally, a feedback control device equipped with an auto-tuning mechanism uses the auto-tuning mechanism to identify (determine) the dynamic characteristics of the controlled process (hereinafter referred to as process dynamic characteristics), and calculates P (proportional band), ■ (integral time) , D (differential time), etc., are automatically adjusted to optimal amounts according to process dynamic characteristics to improve functionality and operability.

By the way, for example, the magazine "Insou" (published by Kogyo Gijutsusha)
(01986, Vat, 29 No. ll (7
) l (As described on pages 1 to 58, the process dynamics of conventional feedback controllers of this type can be improved using the frequency response method, where the control loop is completely opened and a special test signal is added. , an open-loop method such as a transient response method, or a Kalman filter method that uses control system noise instead of adding a test signal.
) method, expert method, and other closed-loop methods.

That is, in the case of the open-loop method, as shown in FIG. 2, during adjustment, based on the switching operation of the switch (1), the control signal from the control unit (3), for example, a PID type, of the control device (2) is replaced. Then, a special test signal from the outside, for example a step signal, is applied to the controlled process (4).

Then, it enters the auto-tuning mechanism of the control device (2) and passes from the process (4) through the operator (5) to the control unit (
The step response signal of process (4) outputted to step 3) is sampled to detect the response of process (4).

At this time, the numerical model of the process dynamics is
The functional form of the output signal SO of the step response of process (4) shown in Fig. 3 and the input signal 5i (=step response) of process (4) shown in Fig. Based on the time constant T and proportional gain K (signal of
lOO% level of output signal So/100 of input signal Si
% level), that is, the parameters of the dynamic characteristics are determined and the dynamic characteristics are identified.

Note that in order to accurately determine the time constant T and gain K, the test signal is actually applied repeatedly and the above-mentioned measurements are repeated to identify the process dynamic characteristics.

Then, based on the identification results of the process dynamic characteristics, the optimal values of the control parameters P, I, and D are determined according to the Ziegzer-Nichozs optimal harammeter determination rule, etc., and each parameter of the control section (3) is determined. P, I, and D are automatically adjusted to the optimal values and the device (2)
The control amount is adjusted to the optimal amount.

Note that after the adjustment of each parameter P, I, and D of the control section (3) is completed, the switch (1) is switched and the control section (3)
The control signal is given to the process (4), and the control signal changes based on the difference between the response output signal of the process (4) and the control target value signal, and the process (4) is feedback-controlled.

In addition, in the case of the closed-loop method, the aforementioned switch (1) is not switched, and the waveform measurement of the noise existing in the process itself or the transient waveform measurement of the step response based on setting or changing the target value is performed in the control state. At the same time, process dynamic characteristics are identified from waveform analysis based on the measurement results.

Then, based on the identification results of the process dynamic characteristics, the optimum values of the control parameters P, I, and D are determined in the same way as in the open-loop method, and the control variables are automatically adjusted to the optimum values.

[Problem to be solved by the invention]

By the way, in the case of the open-loop method, the control loop is completely opened and control is disturbed, and process dynamic characteristics are identified under conditions different from normal control conditions based on target values.

Therefore, in the conventional controlled variable adjustment method using the open-loop method, process dynamic characteristics can be identified with relatively simple calculation processing from an easy-to-measure step response waveform based on a step-like test signal. There is a problem that identification cannot be performed and the accuracy of adjusting the control amount cannot be improved.

On the other hand, in the case of the closed-loop method, it is necessary to perform complex waveform analysis of noise waveforms and transient waveforms of response signals, and identification of process dynamic characteristics requires complicated and extremely time-consuming processing.

Therefore, the conventional controlled variable adjustment method using the closed-loop method allows accurate identification and adjustment without disrupting the control, but it requires extremely complicated processing and the calculation processing configuration becomes extremely complicated. , there is a problem that the control amount cannot be adjusted quickly.

In reality, most feedback control devices use the open loop method to adjust the control amount.

SUMMARY OF THE INVENTION An object of the present invention is to provide a control amount adjustment method for a feedback control device that quickly and accurately identifies process dynamic characteristics in a controlled state and adjusts the control amount using simple arithmetic processing.

[Means to solve the problem]

The means for achieving the above object will be explained below.

The present invention provides a control amount adjustment method for a feedback control device that uses an auto-tuning mechanism to identify the dynamic characteristics of a controlled process and adjust the control amount to an optimal amount. Sampling and detecting the step response of the process based on a change in the target value in a controlled state, substituting the detection result into a moment method approximation equation of the dynamic characteristic, calculating a parameter of the dynamic characteristic, and identifying the dynamic characteristic. The present invention provides a method for adjusting a controlled amount of a feedback control device, which is characterized by the following.

[Effect]

Therefore, based on the target value setting and target value change at the start of control, a step signal is given to the controlled process in the controlled state, and at this time, the response of the controlled process changes in steps.

Then, the step response of the controlled process is detected by sampling, and the detection result is found in 11. Measurement of dynamic characteristics and identification of mathematical model are substituted into the simple method of moments approximation formula described by lζ, and process dynamic characteristics are identified from simple algebraic operations.

〔Example〕

Next, the present invention will be explained in detail with reference to FIG. 1 showing one embodiment thereof.

Figure 1 shows the case where it is applied to a feedback control device that performs PTD control. In the figure, the same symbols as in Figure 2 indicate the same or equivalent parts, and (6) is an auto-tuning section that constitutes an auto-tuning mechanism. It consists of a microcomputer, into which the target value signal and the response signal of the process (4) are input, and the control parameters P, I of the control section (3).

Automatically adjust D. (7) and (8) are the control section (3)
control signal path between and process (4), process (4)
These are two normally closed sampling switches inserted in each response signal path between Repeat closing. (9) is 1
This is a feedback control device having the configuration shown by the dotted line.

Then, let the transfer functions of the Laplace expression of the control unit (3) and the process (4) be Gc(S) and Gp(S), and let the target value be U, and the input signal (=control signal)
If the output signal (=response signal) of process (4) based on (t) is y(t), then the transfer function G(S
) is expressed by the following equation (1).

In addition, as described in 11. Measurement of dynamic characteristics and identification of mathematical model in the above-mentioned "Process Control Handbook", the relationship between the transfer function of the entire control system and the impulse response of the system is as follows: Based on this, the dynamic characteristics of the controlled process can be identified by calculating the parameters of the dynamic characteristics using simple algebraic operations from the method of moments approximation equations described below.

In other words, let the transfer function of the entire control system be G(S) mentioned above, and let the k-th moment be Ak (k=o,...). Therefore, as shown in the following equation (3), k The next moment Ak is related to the parameter included in the second derivative of the transfer function G(S), and the unknown n parameters of the function G(S) can be calculated from the n moments by algebraic calculation and
Adjustment parameters P, Q, and R of the tuning section (6),
Control parameters P and I of the control unit (3).

If there is a relationship between p=p, I-P/Q, and D=R/P between The transfer function Gc(S) is expressed by the following equation (4).

Based on the time constant T and proportional gain K, the transfer function Gp(S) is further expressed as
By substituting equations (4) and (5) into equation (1), P, Q, and R in equation (6) are known parameters, and L, T, and K are unknown parameters of the process dynamic characteristics. The process dynamic characteristics can be evaluated by determining L, T, and K.

By the way, the 0th to 3rd moment A in equation (6).

, AI, A2, and Aa are as shown in equations (7) to 0Q, respectively, based on the calculation of equation (3).

(1) 0th-order Momeninote AO (1) 1st-order Momeyunoto Al ...Formula (9), and in the parameters, T and L are the three linear to tertiary of the above-mentioned (8) to G (1) moment AI, A2.A3
It can be found by algebraic operations.

(1) Based on equation (8), the parameter is
1) Determined from the formula.

(ti) Harameter T is based on (9) and 00 formula,
It can be found from Equation (6).

The next formula 00 is obtained.

α・T+β・T +r=o...
Equation 00 and Equation (6) are obtained from Equation 00.

1iQ Harameter L can be found from the above equation (9).

On the other hand, moment Al, A2. A3 is a process (4) based on setting the target value U for starting control and changing the target value U.
) is determined from the step response of

That is, the impulse response of the transfer function G(S) is expressed as g(t
), the following equation 04 is obtained by positively converting equation (2) into a time domain functional equation.

... From equation 03, the next moment Ak can be found using equation α.

Ak = fr″'t'sg(t)dt
...αXun formula Furthermore, the step function δo-U(t-
τ) (δ0 is the step width, U(t) is the unit function).
t), the following equations 09 and 00 hold between f(t) and g(t).

f(t) = sword δo−U(t−τ)・彊τ)dτ
... α9 equation And if the upper limit of integration of α5 equation is ω, it becomes equivalent to setting 1 (= 0) in equation (3), and in the time domain, the step response f (→ Next moment A.

This is shown by the following equation 04.

Furthermore, the next moment Ak excluding the 0th order is determined by the step response f(t) based on the time differential result of the function H(t)=t・[f←)−f(t)] and the ae formula. , f←
) is shown in the alpha frame format.

δo−Ak= ,/,” δO−tk−g(t) d
t Therefore, moments AI, A2, and Aa are found from equation (g) based on the step response of process (4).

Then, due to the step change in the target value U from 0 to the initial value uO at the start of control, and the step change in the target value U when changing the target value from the initial value UO to the old value, for example, the process (4) of the control state is changed. The response signal becomes a step response signal.

Therefore, step response is detected by detecting the response signal at the time of control start and target value change, and based on the detection result,
Calculate the moments At, A2, and AS from equation (to), and also find the moments AI, A2. All is αη formula, (
By substituting equations 6) and (9'), T and L can be found as parameters through algebraic operations, and the operation of process (4) can be calculated without performing complicated processing such as waveform analysis in the conventional closed-loop method. Characteristics can be identified.

Note that δ0 of the α0 formula can be found from uO−Q and u2−old.

In addition, since no steady-state deviation of feedback control remains,
f←) become [0 and u2, respectively.

Therefore, at the time of starting control and changing the target value, the switches (7) and (8) are opened at the period Tc of the sampling clock based on the control of the auto-tuning section (6).

Process (4) through repeated closing and switch (8)
The step response signal is sent to the tuning section (6) with period T.
sampled at c.

By the way, if the target value Iζ set at the start of control and the target value changed at the time of change are UX, then based on the response signal of process (4), that is, the output signal y(t) and the target value ux, the α frame formula Equation 01 is obtained from

Ak==-! u, f: t''-(:ux -y
(t)]d t,. ...
Then, the tuning unit (6) repeats sampling until the state of y(t)=ux is reached, and at each sampling, the tuning unit (6) calculates the output signal y(tn) at the sampling time tn. , the output signal y(tn-
Based on 1), the algebraic operation of the following equation (1) is performed to obtain the moments Al, A2, and A3 from the zero equation.

Furthermore, the obtained moment AI, A2. The value of AI is substituted into formula A9, formula □□□, and formula (9), and T and L are determined from simple algebraic operations as parameters and the process (4
The dynamic characteristics of the engine are quickly identified.

Then, based on the parameters T and L, Ziegler,
Optimum values for the control parameters P, I, and D are obtained through arithmetic processing according to Nichols' optimal parameter determination rule, etc., and adjustment parameters P and Q are determined according to each optimal value.
, R are output from the tuning section (6) to the control section (3), and the parameters P, 1 . D is automatically adjusted to the optimum value, and the control amount is automatically adjusted to the optimum value.

Therefore, the sampling detection of the step response in process (4) based on the step change in the target value at the time of starting control and when changing the target value, and the Kan formula using the sampling value, the Ql) formula, (6) formula, and (9) The moment of the equation; the process dynamic characteristics in the control state are identified by algebraic calculations of the modulus and approximate equations. process dynamics can be quickly identified.

In the above embodiment, the present invention is applied to a control amount adjustment method of a feedback control device that performs PID control, but it can of course be applied to a control amount adjustment method of a feedback control device that performs various types of control such as PI control. .

That is, even if the control method, transfer function Gc (S, etc. of the control device differs from the embodiments, the control method, transfer function Gc
(s), etc., determine the moment and approximate parameters of the controlled process, and use αυ equation, Q4 equation, (9)
By finding the moment method approximation equation equivalent to Eq.
The control amount can be adjusted in the same manner as in the embodiment.

〔Effect of the invention〕

As described above, according to the control amount adjustment method of the feed bank control device of the present invention, the step response of the controlled process based on the step change in the target value at the time of starting control and when changing the target value is used, and the moment method approximation formula is The dynamic characteristics of the controlled process can be quickly identified in the controlled state through simple algebraic processing, and the dynamic characteristics of the controlled process can be quickly and accurately identified using simple processing without complex processing such as waveform analysis. However, the control amount can be automatically adjusted.

[Brief explanation of the drawing]

1 and 2 are block diagrams of a control amount adjustment method of a feedback control device, in which FIG. 1 is an embodiment of the present invention, FIG. 2 is a conventional example, and FIG. 3 is a step response of a controlled process. FIG. 2 is an explanatory waveform diagram. , (3)... PID control section, (4)... Controlled process, (6)... Auto-tuning nog section.

Claims (1)

[Claims] (1) In a control amount adjustment method for a feedback control device that uses an auto-tuning mechanism to identify the dynamic characteristics of the controlled process and adjust the control amount to the optimal amount, when starting control and changing the target value of control, The step response of the process based on a change in the target value is sampled and detected, and the detection result is substituted into a moment method approximation equation of the dynamic characteristic to calculate the parameter of the dynamic characteristic and identify the dynamic characteristic. A control amount adjustment method for a feedback control device, characterized in that: