JPS61138303A - Process controller using mountaineering method - Google Patents

Abstract

PURPOSE:To apply a mountaineering method by determining a slope formula in accordance with measured values of process variables to manipulate process for climbing by a certain height in a prescribed direction, and setting certain limitations periodically to each process variable. CONSTITUTION:Measured values (x) and (y) of process variables are outputted from detectors 3 and 6. An operator 1 receives process measured values xi and yi and an efficiency variable zi inputted through sampling switch SW1-SW3 and calculates a slop formula z=ax+by+c by the regression calculation using the method of least squares. An operator 2 receives this slope formula (z) through a sampling switch SW4 and receives variables (x) and (y) and limit constants DELTAx-DELTAz inputted from the external and performs the mountaineering calculation and gives manipulated variables to controllers 4 and 7 through sampling switches SW5 and SW6. Controllers 4 and 7 drive operation ends 5 and 8 on a basis of input manipulated variables and display process variables (x) and (y). This operation is repeated at a prescribed period, thereby making mountaineering possible while always searching.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a process control device that uses a hill-climbing method for process calculations.

(Prior Art) Mountain climbing methods in the field of statistics are related to methods for determining under what conditions a slope of a mountain should be climbed when the shape and other conditions described in 1.11 are specified by parameters. be. In the mountain climbing method in the case of mathematical programming, the shape of the mountain is generally given by parameters, and the top of the mountain is often determined by off-line calculation. Furthermore, the maximum slope method is usually used to climb up the slope of a mountain. The maximum slope method is a method of climbing in the direction of the steepest slope of the mountain, in proportion to the slope.

(Problems to be Solved by the Invention) Several problems arise when attempting to apply such a mountain climbing method to the field of process control.

First, as mentioned above, in conventional mathematical planning, etc., the method of finding the top of the mountain is taken by off-line calculation, so it was not considered that it could be applied to actual process control. It will be done. Second, in conventional mountain climbing methods, the main purpose is to find the top of the mountain, and once the top of the mountain is found, the search ends.

In the case of process control (online), even if you are at the top of the mountain at one time, it does not necessarily mean you will be at the top of the mountain at the next time. This is because process variables change from moment to moment.

The present invention has been made in view of these points, and its purpose is to realize a process control device in which the climbing method is applied to a process control system.

(Means for Solving the Problems) The present invention, which solves the above-mentioned problems, uses a statistical method to determine the slope equation at that time from the measured values of a plurality of process variables, and calculates the determined slope equation. The method is characterized in that the process variable is manipulated to climb a certain height in the direction of the slope with the maximum inclination based on the slope, and the operation of applying a certain limit to each process manipulated variable is periodically performed. That is.

(Example) Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention. In the figure, 1 receives process variable measurement values xi, yi and efficiency variable 71 inputted via sampling switches S W 1 to S W s and calculates the slope equation z = ax + by + c at that time using a statistical method. This is the first arithmetic unit. As a statistical method, for example, a least squares method may be used to find a regression equation that closely matches the measurement results. 2 receives the slope equation 2 output from the first arithmetic unit 1 via the sampling switch S W s and is a limiting constant Δ× of the process variables x, y and the efficiency variable l.

This is a second arithmetic unit that receives Δy and ΔZ and performs up-and-down calculations. These first and second arithmetic units 1゜2 are as follows:
For example, a micro combi coater is used.

3 is a detector that detects the process variable x; 4 is the detector 3;
a first controller that receives the output of and the manipulated variable Δx output from the second computing unit 2 via the sampling switch SWs, performs a predetermined P[D computation, and drives the operating end 5;
6 is a detector for detecting the process change fly; 7 is the output of the detector 6 and the manipulated variable Δy output from the second computing unit 2;
This is a second controller that receives the signal via the sampling switch 5WII, performs a predetermined PID calculation, and drives the operating end 8. For example, temperature is used as the process variable x, and pressure, for example, is used as the process variable @y. As is clear from the figure, upper and lower limits of operation input are used for the first and second controllers 4.7, respectively. Further, the sampling switches SWs to SWs are turned on and off in synchronization with each other at a fixed timing. The operation of the device configured as described above will be explained as follows.

From the detectors 3.6, the measured values of the process variables x,
■ is output. These process variables x.

(2) and the efficiency variable 1 are sampled over a certain period of time so long that the dynamic characteristics of the process can be ignored, and are taken into the first computing unit 1. The first arithmetic unit 1 determines the slope equation at a given time based on the acquired data using the statistical method described below. Here, the slope equation is determined by regression calculation using the least squares method. Now, as the formula S, S−Σ(zi −(axi +byi +c ))
2 However, i = 1.2. ..., n is a coefficient a representing the slope equation z-ax+by+c
, b, and c are given by the following equations, respectively. That is, as/la −Σzixi−aΣX12−b 2:xiyi−cΣ×
1=Oas/ab =Σziyi-aΣxiyi-bΣyi2-cΣyi=
Oas/a.

Coefficients a, b, and c can be obtained by solving simultaneous equations for a, b, and c expressed as Σzi−aΣxi−bΣyi−cn=0. Note that the constants Σztxi, Σ×12°Σxtyi, etc. in equation (2) are sequentially stored in the memory in the arithmetic unit 1 for each sampling.
Alternatively, it can be easily obtained by sequentially storing the calculated results.

In this way, when the coefficients a, b, and c are determined, the slope equation z
= ax+by+c has been found. In the present invention, even if the shape of a mountain cannot be specified, the slope equation of the mountain can be determined by using a statistical method. FIG. 2 is a diagram showing the conditions of slopes and mountains determined in this manner. In the figure, Yaga represents a mountain, and B represents a part of the slope.

The slope equation determined by the first calculator 1 is given to the second calculator 2 via the sampling switch SW 4. The second calculator 2 determines the direction of maximum inclination from the slope equation. Now, if we express the slope equation as an equation near the operating point, we get
The slope equation ΔZ near the operating point shown in FIG. 2B can be expressed as ΔZ=aΔx+bΔy (3). Next, set a limit C for ΔX and Δ■ as shown in Figure 2, Δ×2 + ΔV 2 = K (4), and find the maximum value of Δ2 under this limiting condition. direction (direction of mountain θ)
give.

Now, in order to find the maximum value of Δ2, Δ2 ′ −Δ2 +λ (K−Δ× 2−Δy 2
)=a Δx+b Δy +λ (K−Δ× 2−Δy 2 ) and by partially differentiating Δ2′ with Δ× and Δ■, the following equation is obtained.

aΔZ'/eΔx=a −2λΔx −QaΔZ'/a
Δy=b−2λΔy−0 When equation (6) is solved for 2λ, the following equation is obtained.

2λ=a/Δ×=b/ΔV −(a2+h2)/Δ2 From now on, Δx = (a/(a 2+b 2)) Δ2Δv
−(b/(a2+b2))Δ2 In other words, if you move by ΔX and Δy shown in equation (8),
In other words, you have climbed the mountain by a predetermined amount in the direction of maximum inclination. Note that if Δ×, Δytr are given negative signs, the sequence will go down the mountain.

Here, if the climbing speed in one step is ΔZo, then 1
The operation amount for each time is calculated from equation (8) as follows: Δx − (a /
(a 2 +b 2 ))ΔZ.

ΔV − (b / (a 2 + b 2)) ΔZ.

becomes. By the way, according to the present invention, even after reaching the top of the mountain, the search is always performed at a constant cycle. Therefore, depending on the situation, there is a possibility that you will try to climb using the Step of Flash. Therefore, limits are set for each of the amounts of one-time operation of ΔX and ΔV.

The limit amounts ΔxOrΔyO and ΔzO of ΔX and ΔV are given to the second arithmetic unit 2 from the outside, for example, by a key-in operation from a keyboard. As a result, Δx and Δy are respectively given by the following equations.

Here, in equation (10), Signx is the sign of x'' (+ or -
), and 1n (ΔXo, lx l) is the given limit ΔxO and the absolute value lxl of the X coordinate of the operating point,
Indicates that the smaller value is selected. That is, it is sufficient to determine one operation mΔ×, ΔV given by equation (9) under the restriction shown by equation (10). This operation amount Δ×, Δy
are the sampling switches SWs from the second arithmetic unit 2,
It is applied to each controller 4.7 via SW6.

The controller 4.7 drives the operating end 5.8 based on the input operation amounts Δx and ΔV, and displays the process changes lx and y. By repeating such operations at predetermined intervals, it becomes possible to climb the mountain while always searching.

Note that if the coordinates (x, y) of the operating point have restrictions such as upper and lower limits, the operation may be clamped to the restrictions and only variables that are not restricted may be moved under the same conditions. Alternatively, the above operations may be performed except for limited variables.

In this case, the variable is one-dimensional, so the amount of operation is The same can be applied to Further, the statistical method for determining the slope equation is not limited to the above-mentioned regression calculation using the least squares method, and other statistical calculation methods may be used.

(Effects of the Invention) As explained in detail above, according to the present invention, the mountain climbing method can be used for process control, and the slope formula can be obtained even if the shape of the mountain is not known in advance, making it possible to put it into practical use. The above effect is extremely large.

[Brief explanation of drawings]

FIG. 1 is a configuration block diagram showing one embodiment of the present invention, and FIG.
The figure shows mountains and slopes. 1.2... Arithmetic unit 3.6... Detector 4.7.
...Controller 5.8...Operation end SW 1~S
'A' e...Sampling switch X>N

Claims (1)

[Claims] The slope equation at that time is determined using statistical methods from the measured values of multiple process variables, and the process variable is set to climb a certain height in the direction of the slope with the maximum slope based on the determined slope equation. 1. A process control device using a hill-climbing method, characterized in that the device is configured to periodically perform an operation in which a certain limit is applied to each process operation amount.