JPS62175805A - Adaptive control system - Google Patents

Abstract

PURPOSE:To attain the optimum process control at all times regardless of the change of process characteristics by estimating the change of the process characteristics only when needed and performing usually the optimization of the control parameter. CONSTITUTION:An optimum control parameter arithmetic part 3 calculates an optimum parameter based on the identification data collected by a data collecting part 2. A bidirectional adaptive control learning part 4 chooses the adaptive exponent of an object process based on said optimum parameter. This adaptive exponent is fixed and the process characteristics are estimated. Based on the result of this estimation, the optimum control value is obtained and stored in a pair with said adaptive exponent. These operations are repeated to obtain a parameter gathering against the entire operation of a process. Thus an approximation curve is obtained for the stored data. These processes to collect data and obtain the approximation curve are carried out only when needed. While a PID control output arithmetic part 1 obtains the optimum parameter from an obtained adaptation curve and performs the control arithmetic operation by means of said optimum parameter.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an adaptive control method that responds to changes in process characteristics in closed-loop control of a process.

[Conventional technology]

As one of the conventional parameter adaptation methods of this type, the fourth
The one shown in the figure is known. This feeds back the output amount pv of the process 5 to the adjustment section 6, and outputs the output amount M of the adjustment section 6.
In performing closed loop control of process 5 using V,
In addition to investigating in advance how the operational characteristics of Process 5 will change and clarifying the process quantity (adaptation index) CV, which has a deep relationship with changes in the characteristics, as an indicator of changes in process characteristics, When the characteristics change, it is also clarified how the control parameters should be changed to adapt to the change in characteristics, and based on the process quantity (adaptation index) CV, which is an index of the change in process characteristics measured for process 5. Adaptation signal A
The adaptation signal AV is generated by the adaptation device 7, and the adaptation signal AV is given to the adjustment section 6 to change the control parameters.

However, according to such a parameter adaptation method, the relationship between the characteristics of the process 5 and the process quantity C, which is an index of its change, changes from the adaptation curve A to the dotted line B based on previously investigated data shown by the solid line in FIG. If there is a deviation as shown in , there is a drawback that the control parameters cannot be correctly adapted to changes in process characteristics.

On the other hand, the one shown in FIG. 6 is also known.

This is when performing closed loop control of process 5.
The state of the operating characteristics of the process 5 is constantly checked, and the output amount PV of the process 5 and the output amount MV of the adjustment section 6 are transmitted to the adaptation device 7 using the configuration shown in the figure. Based on the relationship between input and output of process 5, process characteristics are estimated to accurately grasp the operational characteristics of process 5, and control parameters are applied to the moment-by-moment process characteristics revealed by the estimation. In this method, an adaptation signal AV for adapting the process is supplied to the adjustment section 6, and the control parameters are changed to adapt to the momentary state of the process.

However, according to this parameter adaptation method, when the process characteristics change, the adaptation device 7 uses data processing to estimate new process characteristics and grasp them correctly, but this task takes time. Therefore, there is a drawback that the control parameters cannot be changed immediately in response to new states of process characteristics.

[Problem that the invention seeks to solve]

In other words, there are two issues in adjusting control parameters in process control: (a) determination of optimal control parameters for fixed process characteristics; and (u) conventional method of adapting control parameters to follow changes in process characteristics. However, there is a problem in that it is not possible to satisfy these two issues at the same time. Therefore, an object of the present invention is to provide an adaptive control method (hereinafter also referred to as a two-way adaptive control method) that realizes "estimation of process characteristic changes" and "optimization of control parameters."

Means for solving problem c] Control output that calculates the optimal control parameters corresponding to the operating conditions using an adaptation curve given as a function of the adaptation index in order to cope with the nonlinear characteristics of the process, and performs control calculations using these parameters. a calculation means, a data recording means for determining process characteristics corresponding to different adaptation indices, and a calculation means for calculating optimal control parameters from the obtained data;
A learning means is provided which determines the validity of the newly determined value from the set of control parameters that have already been determined, and if it is valid, adds the new value to the set of control parameters to update the adaptation curve.

[Effect]

Control calculations are always executed by the control output calculation means to deal with fluctuations in process characteristics due to changes in operating conditions, and data recording, optimum control parameter calculations, and learning of adaptation curves are performed as appropriate, resulting in very slow operation. This makes it possible to cope with fluctuations in process characteristics due to aging.

〔Example〕

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

In the figure, 1 is a control output calculation section, 2 is a data collection section, 3 is an optimal control parameter calculation section, and 4 is a learning section.

In other words, in order to realize bidirectional adaptive control that combines "estimation of process characteristic changes" and "optimization of control parameters," it is necessary to organically combine the following four functions. .

(1) The control output calculation unit 1 calculates optimal control parameters (PID A speed type PID control calculation method is adopted for the controller so that bumpless adaptation can be performed to these parameters.

(2) The data collection unit 2 collects identification data in order to identify processes. There are two methods for identifying this process parameter: a method using a test signal and a method of determining it from a steady balance state.

(i) Identification using test signals For plants where test signals can be input, test signals are applied to the process and process parameters are identified from the input/output relationship of the process.

(ii) For plants where it is sufficient to adapt only the identified gain due to steady balance conditions,
It is determined whether the system is in a steady state during operation, and the process gain is estimated from the relationship between the manipulated variable and the process value in the steady state of balance.

(3) The optimal control parameter calculation unit 3 calculates optimal control parameters for the process characteristics determined in (2). Many methods have been proposed for this calculation method, but for example, for PID control, the following evaluation function J is determined, (r: set value, X: process value, β: constant)...
-(1) The one that minimizes this can be found using the auto-tuning method.

(4) The learning unit 4 creates and updates an adaptation curve, specifically as follows.

(a) First, what is to be adopted as the adaptation index of the target process is selected.

(b) Next, the process characteristics are estimated from this process adaptation index, and the optimum adjustment value is determined from the estimation result and stored in pairs with the characteristic index.

(c) storing a set of parameters for the entire operating range of the process;

(d) Find an approximate curve of the stored data, that is, an adaptation curve, by applying the method of least squares or the like.

(e) The number of stored data increases as the operating time increases, so when a certain number is exceeded, old data is discarded and new data is stored.

In order to achieve the above, the function (1) is simple to process but requires a high-speed response, and (2)
The functions of (3) and (4) are to deal with changes in the process over time, so the response may be slow, but since it is necessary to manage a large amount of data and perform calculations, it is necessary to use the hierarchy shown in Figure 1. The system is designed such that only function (1) normally operates, and functions (2), (3), and (4) operate when necessary.

The control procedure will be explained with reference to FIG. 2 and FIG. 2A. In addition, FIG. 2 is a flowchart for explaining the control procedure according to the present invention, and FIG. 2A is an adaptation curve (G-
It is a graph showing S curve).

First, an adaptation index for the target process is selected, and the process characteristics are estimated by fixing the adaptation index. The optimum adjustment value is obtained from this estimation result and stored in pairs with the adaptation index. This operation is repeated to store a set of parameters for the entire operation of the process. An approximate curve of the stored data is determined and adaptive control is performed according to this curve. The number of stored data increases as the operating time increases, so when a certain number of data is exceeded, the old data is discarded and new data is stored to cope with changes over time.

In order to make bidirectional adaptive control effective, it is important to accumulate data for creating an adaptation curve that matches the actual process, as shown in FIGS. 2 and 2A. In this regard, one of the points emphasized in the present invention is that it is important to utilize the knowledge of field engineers. for that,
As shown in Figure 3, the data is processed, the optimal PID calculation value is displayed on the screen, its validity is confirmed by comparing it with old data, and the operator can decide whether to use the obtained result or not. It can be operated in a man-machine interactive manner. At this time, the identification method, optimal PI
Regarding the calculation method of the D parameter, the method suitable for the actual process is described in (2) above.

It is possible to select from among the methods described in section (3).

As described above, the speed of adaptation is slowed down as a result of being performed in an interactive manner, but this is not a particular problem since this is an operation to accommodate changes in characteristics due to aging. Usually, the set of pairs of optimal adjustment values and adaptation indexes obtained by this method is calculated by the least squares method as regression coefficients of an adaptation curve with the adaptation index as a variable, and this is used to calculate the nonlinearity of the process. This allows adaptive control to be applied to fluctuations caused by changes in operating conditions.

〔Effect of the invention〕

According to the present invention, it is possible to determine process control parameters that can immediately adapt to changes in process characteristics due to changes in operating conditions, and can also follow and adapt to changes in process characteristics due to changes over time. This provides the advantage that optimal process control is always possible regardless of changes in process characteristics.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram showing an embodiment of the present invention, FIG. 2 is a flowchart for explaining the control procedure according to the present invention, and FIG.
Fig. A is a graph showing an adaptation curve, Fig. 3 is a flowchart for explaining a specific example of the operation of creating and updating an adaptation curve according to the present invention, Fig. 4 is a block diagram showing a conventional example of an adaptive control method, FIG. 5 is a characteristic curve diagram showing a mode of parameter adaptation, and FIG. 6 is a block diagram showing another conventional example of an adaptive control system. Symbol explanation ■... Control output calculation section, 2... Data collection section, 3.
...optimal control parameter calculation section, 4...learning section, 5.
. . . Process, 6. Adjustment unit, 7. Adaptation device. Agent Patent Attorney Akio Namiki Agent Patent Attorney Kiyoshi Matsuzaki Figure 1 Figure 3 Figure 4 Figure 5

Claims (1)

[Scope of Claims] Control output calculation means for determining optimal control parameters from an adaptation curve given as a function of an adaptation index in order to correspond to the nonlinear characteristics of the process, and performing control calculations using the optimal control parameters;
A data recording means that collects data to determine process characteristics corresponding to different adaptation indices, a calculation means that calculates optimal control parameters from the obtained data, and a new set of control parameters that have already been determined. learning means for determining the validity of the obtained value and, if it is valid, adding it to a new set of control parameters to update the adaptation curve; the control calculation by the control output calculation means is always executed; An adaptive control method characterized in that the data collection, optimum control parameter calculation, and learning of the adaptation curve are executed appropriately to update the adaptation curve.