JPH0414101A - Hybrid process controller - Google Patents

Abstract

PURPOSE:To shorten the time required for adjustment of the gain by attaining the automatic readjustment for the gain of a linear manipulated variable and a fuzzy manipulated variable of the fuzzy output gain in accordance with the control performance of a hybrid constitution that is evaluated by a control performance evaluating part. CONSTITUTION:A gain adjuster 38 includes a control performance evaluating part 40, a PI gain adjusting part 42, and a fuzzy output gain adjusting part 44. The part 40 has a US/UR arithmetic part 46 being the ratio of for the manipulated variables of both adjustment and start states, and outputs the ratio value K to the part 44. The part 44 multiplies the present fuzzy output gain Kf by K and adjusts the gain Kf again. The part 42 adjusts the proportional gain Kp in accordance with the control performance evaluated by the part 40. In this adjustment procedure, the requested characteristic is previously acquired and a changed variable is set equal to the precedent value as long as the code of the changed variable of the gain Kp is equal to the precedent one. However the changed variable is set at 1/2 if the code of the value of the gain Kp is adverse to the precedent one.

Description

Detailed Description of the Invention (Field of Industrial Application) The present invention relates to a control device that performs process control such as temperature adjustment control, and particularly to a hybrid process control device that includes linear control such as PI sub-control and fuzzy control. It is.

(Prior art) As a process control device that performs process control such as temperature adjustment control, a P control device or a PI control device uses a control device based on a control deviation between an arbitrarily determined control target value and a control amount from a controlled object. Process control devices of the type that determine manipulated variables, including at least proportional manipulated variables, are well known in the art. This process control device has linear control characteristics and is used for constant value control to maintain temperature and the like at a predetermined control target value.

These control characteristics largely depend on the gain of the manipulated variable, and in order to obtain excellent control performance, it is necessary to accurately adjust the gain.

In response to this, as for conventional linear control devices, a gain adjuster using a limit sensitivity method is known, and this makes it possible to automatically adjust the gain.

(Problem to be Solved by the Invention) A gain adjuster using the limit sensitivity method is effective in a linear control device, but it is not suitable for a hybrid process control device that includes linear control and fuzzy control. lacks effectiveness.

For this reason, in a hybrid process control device, various gains for linear control and various parameters for fuzzy control are individually adjusted, and then each gain is readjusted in the hybrid configuration. Regarding the readjustment of the gain in this hybrid configuration, since the adjustment procedure is not clear, it has to be carried out by trial and error, and there is a problem in that it cannot be automated.

The present invention has been made by focusing on the above-mentioned problems in the conventional hybrid process control device as described above, and is a hybrid process that automates gain readjustment and eliminates the need for troublesome gain readjustment work. The purpose is to provide a control device.

(Means for Solving the Problems) According to the present invention, the above object is to provide a linear manipulated variable determining means for determining a linear manipulated variable including a proportional manipulated variable based on a control deviation between a control target value and a controlled variable. and a fuzzy manipulated variable determining means that determines a manipulated variable by fuzzy inference according to the control deviation, a control performance evaluation unit that evaluates control performance of a control system, and a control performance evaluation unit that evaluates control performance of a control system, a gain adjustment unit that adjusts the gain of the linear manipulated variable according to the control performance evaluated by the evaluation unit; and a fuzzy controller that adjusts the output gain of the fuzzy manipulated variable according to the control performance evaluated by the control performance evaluation unit. This is achieved by a hybrid process control device characterized by having an output gain adjustment section.

(Operation) According to the above configuration, the gain of the linear manipulated variable and the fuzzy output gain fuzzy manipulated variable are automatically readjusted according to the control performance in the hybrid configuration evaluated by the control performance evaluation section.

(Example) The present invention will be described in detail below with reference to the accompanying drawings.

FIG. 1 shows the basic configuration of a hybrid process control device according to the present invention. The hybrid process control device according to the present invention includes a proportional manipulated variable calculating section 10 and an integral manipulated variable determining means 12 as linear manipulated variable determining means, and a fuzzy inference section 14 as a fuzzy manipulated variable determining means.

The proportional operation amount calculation unit 10 calculates the control amount Y from the controlled object 16.
A control deviation e between the control target value R and the control target value R is added at a point 18, and a proportional operation amount is calculated based on this control deviation e.
The amount of repetition is output to the proportional gain setting device 20.

The proportional gain setter 20 has a proportional gain Kp determined by a gain adjuster 38, which will be described later, and a proportional operation amount calculation unit 10.
By multiplying the manipulated variable by the proportional gain Kp, a proportional manipulated variable Up is generated, and this is output to the summing point 26.

The integral operation amount calculation unit 12 calculates the control amount Y from the controlled object 16.
An integrator 22 calculates the integral value of the control deviation e between the control target j and the control target
The integral operation amount is calculated based on this integral value, and the operation amount is output to the integral gain setting device 24.

The integral gain setter 24 is set with a predetermined integral gain Ki, and generates an integral manipulated variable Ui by multiplying the manipulated variable from the integral manipulated variable calculating section 12 by the integral gain Ki;
This is output to a summing point 26.

The proportional manipulated variable up and the integral manipulated variable Ui are added at a summing point 26, and a linear manipulated variable Upi consisting of the maximal value 1 is outputted to the calculation means 30 with a ratio.

The fuzzy inference unit 14 calculates the control deviation e from the addition point 18.
is given, fuzzy inference is performed in accordance with the fuzzy rules as shown in FIG.

In the fuzzy rule shown in Figure 2, NL, NS, ZR, PS, and PL are used as linguistic information, where NL is negatively large, NS is negatively small, and ZR is approximately These are fuzzy labels indicating zero, PS is positively small, and PL is positively large.

The fuzzy gain setter 28 is a gain adjuster 38 which will be described later.
The fuzzy output gain Kf is determined by the fuzzy output gain Kf and the fuzzy inference output f is multiplied to generate the fuzzy manipulated variable Uf, which is then added to the ratio by the calculating means 32.
It is designed to output to.

The ratio calculation means 30 and 32 give the linear control ratio (linear control weighting is a coefficient) α and the fuzzy control ratio (fuzzy control weighting is a coefficient) 1−α determined by the manipulated variable ratio changing means 34. The calculation means 30 multiplies the linear operation amount Upi from the summation point gain 26 by the linear control ratio α, and the weighting is calculated by calculating the linear operation amount α・Upi.
The calculation means 32 multiplies the fuzzy operation amount Uf from the fuzzy gain setter 28 by the fuzzy control ratio ¥-1-a, and the weighting is performed using the fuzzy after calculation. Operation @(1-α) Add Uf to point 3
It is designed to output to 6.

The addition point 36 is weighted by the linear operation amount α・U after calculation.
pi and weighting are fuzzy manipulated variables (1-α) Uf after calculation
are added to each other, and the total operation amount U is made up of the total value.
is given to the controlled object 16.

The operation amount ratio changing means 34 is given information regarding the control deviation e and the degree of adaptation of e=ZR from the fuzzy inference unit 12, and accordingly, when the control deviation e is large, α is made smaller than when it is small, On the other hand, when the control deviation e is small, α is made larger than when it is large.

As a result, when the control deviation e is large, that is, at startup, the ratio of the fuzzy manipulated variable Uf to the determination of the total manipulated variable U becomes large, and the ratio that the linear operation 1tUpi contributes to the determination of the total manipulated variable U becomes small; When the control deviation e is small, that is, when it is settled, the ratio of the fuzzy manipulated variable Uf to the determination of the overall operation MU becomes small, and the linear manipulated variable Up.
The ratio that i gives to the determination of the total manipulated variable U increases. Therefore, during startup, control is performed primarily by the fuzzy manipulated variable Uf, and during settling, control is primarily performed by the linear manipulated variable Upi.

The gain adjuster 38, as best shown in FIG.
A control performance evaluation section 40 evaluates the control performance of the control system given the control target value R, the total operation amount U, and the control amount Y, and a proportional gain Kp according to the control performance evaluated by the control performance evaluation section 40. and a fuzzy output gain adjustment section 44 that adjusts the fuzzy output gain Kf according to the control performance evaluated by the control performance evaluation section 40.

As clearly shown in FIG. 4, the control performance evaluation unit 40 calculates the ratio K between the operating amount US during settling and the operating amount UR during startup.
It has a US/SR calculation unit 46 that calculates K=U
The value of S/UR is output to the fuzzy output gain adjustment section 44.

Incidentally, when the operation ff1Us at the time of settling is oscillating, it is sufficient to find the average operation amount.

The fuzzy output gain adjustment section 44 multiplies the current fuzzy output gain Kf by , that is, Kf=X-Kf.
The fuzzy output gain Kf is readjusted by performing the following calculation. This readjustment is US (!
: It only needs to be done once if the value is significantly different from the UR.

As shown in FIG. 4, the control performance evaluation section 40 includes an overshoot measurement section 48 for proportional gain adjustment, a target value achievement time measurement section 50, and a hunting frequency measurement section 52.
Information O8 representing the magnitude of overshoot, information ST representing the target value achievement time, and information HN representing the number of hunting times are provided to the PI gain adjustment unit 42.

As a method for adjusting the proportional gain Kp, since overshoot and target value achievement time have an antinomic relationship with each other, the following method may be considered.

■Give the required characteristics in advance.

■The amount of change in the value of proportional gain Kp is the same amount of change as last time if the sign of the amount of change in proportional gain Kp is the same as last time,
When the sign is reversed from the previous time, the amount of change is halved.

FIG. 5 shows an example of a method for adjusting the proportional gain Kp. In this case, first, in step 10, the magnitude of the overshoot is determined. If the overshoot is large, the process proceeds to step 40; otherwise, the process proceeds to step 2.
Go to 0.

In step 20, the number of huntings is determined. If the number of huntings is large, the process proceeds to step 40; otherwise, the process proceeds to step 30.

In step 30, a determination is made as to whether the time required to reach the target value is large or small. If the time required to reach the target value is large, the process proceeds to step 50; otherwise, this routine ends.

Step 40 is executed when the overshoot is large or the number of hunting is large, and in this step 40, the proportional gain Kp is reduced.

Step 50 is executed when the overshoot is small, the number of hunting is small, and the time required to reach the target value is long, and in step 50, the proportional gain Kp is increased.

Next, an example of an actual adjustment procedure will be explained.

As an example of the parameter adjustment procedure, the following control target is assumed. This control target is a model that approximates a water heater with 1-order lag + dead time, and the maximum water input amount is 9 liters, the minimum operation amount is 4000 kg, and it is identified from the step response of force lorry/hour. The function Gp+ (S) is expressed as follows.

Gp+ (s) = f9.9 / (1+3.67s
) l e-"Here, the required specifications are an overshoot of 10° C. or less, a hunting frequency of 3 or less, and a target value achievement time of 30 seconds or less.

First, set the initial setting gain to another control target (maximum water input amount 9
K p =0.25, K i =10.8
, K f =1.0, and a target temperature of 75°
The simulation results when C is shown in FIG.

From the simulation results shown in Fig. 6, the manipulated variable UR at the time of startup is almost equal to the average manipulated variable when the controlled variable is oscillating around the control target value, so the control performance evaluation section calculates the fuzzy output gain Kf. Information on overshoot, hunting frequency, and target achievement time is given to the PI gain adjustment section without changing the values.

In the PI gain adjustment section, the Kp gain is multiplied by 1/2 since the open neat temperature is as large as 27°C.

Therefore, set K p =0.125 and set the target temperature to 75°C again.
operate as. The simulation results at this time are shown in FIG.

Based on the simulation results shown in FIG. 7, the control performance evaluation section newly provides information regarding the overshoot, the number of /X synchronizations, and the target value achievement time to the PI gain adjustment section.

In the PI gain reduction section, although the overshoot has decreased from the previous time, it is still large at 243℃, and the Kp
Further multiply by 1/2. This results in K p =0.06
Then, the target temperature is set to 75°C and the operation is started again. The simulation results at this time are shown in Figure 8.

Based on the simulation results shown in FIG. 8, the control performance evaluation section provides information regarding open neatness, hunting frequency, and target value achievement time to the 19PI gain adjustment section.

In the PI gain adjustment section, overshoot -0.1℃
, the number of hunting is 1 time, or the target value achievement time is 80
Since the time is as large as seconds, the Kp gain is multiplied by 1.5. Therefore, set K p =0.09 and set the target temperature to 75° again.
Operate as C. The simulation results at this time are shown in FIG. Further, the simulation results focusing on the control amount near the target value in FIG. 9 are shown in FIG. 10.

From the simulation results shown in FIGS. 9 and 10, the control performance evaluation section again provides information regarding overshoot, hunting frequency, and target value achievement time to the PI gain section.

As a result, the PI gain adjustment section has an overshoot of 0.8
°C, the number of hunting times is 2, and the target value achievement time is 22 seconds, which satisfy all the required specifications, so the gain adjustment is completed.

(Effects of the Invention) As understood from the above explanation, according to the hybrid process control device according to the present invention, the linear operation amount gain and fuzzy The output gain and the fuzzy manipulated variable are automatically readjusted, eliminating the need for troublesome gain readjustment work and shortening the time required for gain adjustment.

[Brief explanation of drawings]

FIG. 1 is a schematic configuration diagram showing one embodiment of the hybrid process control device according to the present invention, FIG. 2 is a rule table diagram showing an example of fuzzy rules used in the hybrid process control device according to the present invention, and FIG. A block diagram showing the specific configuration of the gain adjuster in the hybrid process control device according to the present invention, and FIG. 4 is a block diagram showing the specific configuration of the control performance evaluation section in the hybrid process control device according to the present invention. , FIG. 5 is a flowchart showing the proportional gain adjustment routine in the hybrid process control device according to the present invention, and FIGS.
FIG. 0 is a time-series graph showing simulation results of parameter adjustment. 10... Proportional manipulated variable calculating means 12... Integral manipulated variable calculating means 14... Fuzzy inference section 20... Proportional gain setting device 24... Integral gain setting device 28... Fuzzy output gain setting device 38...Gain adjuster 40...Control performance evaluation unit 42...PI gain adjustment unit 44...Fuzzy output gain adjuster Patent applicant: Omron Corporation Representative Patent attorney Kazu 1n Seirori No. Figure Figure 1 Missing 1 1st Tree 1 State 8 Lottery 8 +2.3 Akira 2 Figure 1 Figure 1 1st N1] IM+ Procedural Neho Seisho (self-proposal) August 24, 1990 , g3g3E3゜+2.0 people ρ 2゜3゜4゜Patent Application No. 118047 No. 1997 Name of the invention Relationship to the case of person who corrects hybrid process control equipment Patent applicant Address 10, Hanazono Tsuchido-cho, Ukyo-ku, Kyoto City Name
(294) Yoshio Tateishi, Representative of Omron Corporation

Claims (1)

[Scope of Claims] 1. Linear manipulated variable determining means for determining a linear manipulated variable including a proportional manipulated variable based on a control deviation between a control target value and a controlled variable;
In a hybrid process control device having a fuzzy manipulated variable determining means that determines a manipulated variable by fuzzy inference according to the control deviation, a control performance evaluation unit that evaluates control performance of a control system; and the control performance evaluation unit a gain adjustment section that adjusts the gain of the linear manipulated variable according to the control performance evaluated by the control performance evaluation section; and a fuzzy output gain that adjusts the output gain of the fuzzy manipulated variable according to the control performance evaluated by the control performance evaluation section. A hybrid process control device comprising: an adjustment section;