JPS59139404A - Pid control method - Google Patents

Abstract

PURPOSE:To reduce the cost and to operate a process adaptively to even disturbance flexibly by calculating a forgetting factor, which is a parameter determining availability, on a basis of an identification error in an identifying part which identifies the process. CONSTITUTION:In a PID controller 6 of the process which controls automatically the temperature and etc. in a plant, a deviation e(t) between a control target r(t) and a control signal y(t) outputted from a process 1 is operated in an operating part 5. An operation signal u(t) of the process 1 is generated from a manipulated variable operating part 2 on a basis of this deviation, and PID parameters are adjusted in an identifying part 3 and a control parameter operating part 4. Though the identifying part 3 uses a preliminarily given forgetting factor to operate the identification quantity of the process, this forgetting factor is adjusted at every synchronous gamma by sampling control, and an information quantity attained by weighing the quantity, which indicates an identification error at a time (t+gamma), with parameters including the forgetting factor at a time (t) is used for this adjustment.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a DDC (Direct) igi using a microcomputer (hereinafter simply referred to as a microcomputer).
The present invention relates to a PID control method having a function of tuning PID control parameters necessary for control in a PID control system.

[Prior art]

As a control method using conventional PID parameter tuning, for example, a sample value PID control device (Japanese Patent Laid-open No. 5
The method used in Japanese Patent No. 7-23108 requires complicated calculations such as solving the roots of cubic equations, and also requires greater calculation speed and number of calculations.
The drawback was that it could not be run on a small microcontroller.

In addition, the combination of the identification method based on step-like signal responses and the design method of closed-loop specified PID parameters (see the Institute of Electrical Engineers of Japan Study Group Materials '82. System and Control Study Group edition) requires complicated calculations such as Fourier transform. Therefore, it had the same drawbacks as above.

[Purpose of the invention]

An object of the present invention is to solve the above-mentioned drawbacks of the conventional method, and to make it practical even with a small microcomputer, (1) the calculation capacity and number of operations can be reduced. (2) It can flexibly respond to changes in target set values and disturbances, whether sudden or gradual. An object of the present invention is to provide a PID control method having a PID auto-tuning function that satisfies the following two requirements.

[Summary of the invention]

In order to achieve the above object, the present invention uses a normal identification method in the identification section that sequentially identifies processes from each time series data of operation signals and control signals, and forgets past information based on the size of identification error. By automatically calculating the coefficient (hereinafter referred to as forgetting factor), the above requirement (2) is made possible, and the identification method having the above forgetting factor and the calculation of discrete PID parameters in the control parameter calculation section are used. possible])e
ad-beat type PID parameter calculation method (Auto
matica1980, Vol 16 PP117~
133) is used to simplify calculations and enable requirement (1) above.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of a PID control device to which the present invention is applied. The object to be controlled is, for example, a process 1 in which temperature, pressure, flow rate, etc. in a steel or chemical industrial plant are automatically controlled.

The PID control device is part 6 in FIG. 1 and has the following configuration. Deviation e(t) between control target γ(1) and control signal y(t) output from process 1 (e(t)=
γ(t)−y(t)) and the deviation e
(t) to generate the operation signal 13 (t) to the process 1 according to the PID control parameters, and each time series data of the control signal y (t) and the operation signal 1) and the pulse of the process 1. Identification unit 3 that identifies coefficients of transfer function
From the identified coefficients 9<,>, the PID control constant K c , 'l' I which is the control parameter of the manipulated variable calculation unit 2
.

A PID control device 6 is configured by a control parameter calculation unit 4 that calculates Td. That is, this configuration consists of basic feedback control and elements that adjust PID parameters in accordance with process variations.

Next, each component and operation of the PID control device 6 will be explained in more detail. The manipulated variable calculation unit 2 that controls the process 1 uses PID control constants Kc and Tt, which will be described later.
, Td, the following algorithm is processed from the time series data e (t), and the operation signal u(1) is input to the process.

・・・・・・・・・(1) u(t)=u(t-τ)+Δu(t)・
・・・・・・・・・(2) Equations (1) and 9 Equations (2) are well-known speed-type PID algorithms, where KC is a proportional gain,
TI is an integral time constant, Td is a differential time constant, and τ is a sample control period.

The operation of this manipulated variable calculating section 2 is shown in FIG. Next, we will explain the identification unit 3 that estimates process parameters from the output y (k) of process 1 obtained after inputting the operation signal U (g) to the process. Explain the model.

An equivalent model (pulse transfer function) of the process is obtained from the operation signal u (t) and the control signal y (t) as follows.

Here, the delay time equivalent to the first-order delay element zi is the sample control period τ, and d is the process delay time. Also, the unknown parameter al to be identified.

Gold “＝〔Incense・、◇・、・・・、zen・、wo、、rei・、・・・
. . all .] . . . (4) The calculation procedure of the identification section 3 is shown in FIG. The algorithm of the identification unit 3 usually performs calculations based on the method of least squares and recursive calculations, and can be expressed as follows.

Rei(1+te)-Tai t) 10F(t)Z(t-Moτ)Cy(
t + τ) - z" (plan f)]...
(5) ・・・・・・・・・(6) Here, F (t + r ) is (n+m)x (n+m
), and Z is an observation vector expressed by the following equation.

Z (t+r)=C-y(t),-y(t-r),...
-y(t-nr).

u (t-d r), u (t-r-d r),
・-・, u (t-mτ-d r) )...
...(7) Also, ρ(t+r) is the forgetting coefficient (forgetti
ngfactor ), and in the conventional example, it is selected as follows. If the forgetting coefficient, which is an important factor determining the effectiveness of various identification problems, is determined as described above, it cannot adequately cope with the dynamic characteristics of various processes. For example, if α is chosen close to 1, tracking may be delayed if the process characteristics change rapidly, or conversely, if α is chosen to be a small value, the variation in identified values may be large due to external disturbances, etc. However, there was a problem in that it was not possible to perform a good identification. Therefore, the identification error ε(t
)=y(t) −Z(t) (−τ), the forgetting coefficient is automatically adjusted as follows. The method is shown in the following equation.

[y(t+τ)−zT(10τ bath(t):]20≦ρ(
t+τ)≦1 (8) Here, Σ, = (expected value of identification error variance) × (expected value of regression period) The second term on the right side of equation (8) The meaning is that it indicates the ratio of the amount of information obtained at time (1+τ), so ρ(
t+τ) indicates the rate at which the amount of past information is retained without being forgotten. For example, g(t+τ)=[y(t+r)−ZT
If (t+r)'2ゝ(heavy)] becomes larger, (ten)
The amount of information obtained at any given time is large, and a large amount of past information is also forgotten. By sequentially determining the forgetting coefficient in this way, it is possible to know how reliable the currently obtained information is compared to past information, making it possible to perform flexible adaptive operations. If we express the pulse transfer function of the process using the identified parameters, we get:

Next, from the identified coefficient G(t), the PID parameter Ke,
The algorithm procedure for the control parameter calculation unit 4 that determines TI and Td will be described.

Here 0〈r1≦1 ・・・・・・・・・(12) ・・・・・・・・・(13) qz = q6 (u”(t)−δ−) ・・・・・・
...(14) τ Q+=δ-qOQ2 Rabbit...(15)
Ke=q.

Td=- q.

The above algorithm is summarized in a flowchart and shown in FIG.

〔Effect of the invention〕

As explained above, according to the present invention, the forgetting coefficient, which is an important parameter that determines the effectiveness of various identification problems, is calculated in the identification unit that identifies the process based on the identification error, so that the target setting value can be adjusted. This has the effect of making it possible to perform flexible adaptive operations regardless of whether changes or disturbances are sudden or gradual. In addition, by combining the identification method that produces the above effects and the method of calculating easily-calculated discrete PID parameters in the PID parameter calculation section, it is possible to reduce the memory capacity of the digital controller and the number of calculations, so the system as a whole , it is possible to simplify the configuration and reduce the computer cost associated with it.

[Brief explanation of the drawing]

FIG. 1 is a block configuration diagram of a PID control device to which the present invention is applied, FIG. 2 is a flowchart of the manipulated variable calculation section, FIG. 3 is a flowchart of the identification section, and FIG. 4 is a control parameter calculation section. This is a flowchart. 1... Process, 2... Manipulated amount calculation section, 3... Identification section, χ 1 Figure Y 2 Figure χ 3 Figure

Claims (1)

[Claims] 1. A deviation amount calculation unit that calculates the deviation amount e(1) between the control signal y(1) output from a PID-controlled process and the target value signal γ(1) of the process; , and a control parameter calculation unit that calculates the e(t) and inputs it to the manipulated variable calculation unit, wherein the forgetting coefficient is adjusted every sample control period τ.
D control method. 2. The forgetting coefficient ρ (in 10) at time (t + τ) is a quantity representing the size of the identification error at time (t + τ),
2. The PID control method according to claim 1, wherein the PID control method is determined based on the amount of information obtained by weighting the amount including the forgetting coefficient ρ0 at time t. 3. The PID control method according to claim 1, wherein the forgetting coefficient ρ (10τ) is determined by equation (8).