JPS63113736A - Method for parallel running of plural fuzzy inference arithmetic unit - Google Patents

Abstract

PURPOSE:To ensure common use of a consequent part computing element with plural antecedent part computing elements by selecting the largest grade value via said consequent part computing element after having comparison between the grade values corresponding to the same consequent part fuzzy variable among those grade values read out of the grade memories of each antecedent part computing element and then carrying out a consequent part arithmetic operation of a control rule only with said largest grade value. CONSTITUTION:A fuzzy inference computing element is divided into antecedent part computing elements 10a and 10b which perform only antecedent part arithmetic operations and a consequent part computing element 20 which carries out only a consequent part arithmetic operation. That is, just a single element 20 is used in common with two elements 10a and 10b. The element 20 compares the grade values corresponding to the same consequent fuzzy variable with each other to select the maximum value among those grade values read out of elements 10a and 10b. Then a consequent part arithmetic operation of a control rule is carried out only with said maximum value. In such a way, the cost resuction is possible with parallel operations of plural fuzzy inference arithmetic units.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a fuzzy inference calculation device for fuzzy adjustment performed by applying so-called “fuzzy logic”, and more specifically, to The present invention relates to a method for operating multiple units in parallel.

Assume that in a fuzzy inference arithmetic device, if the number of control rules used in the device is 10, the number is fixed as 10. However, if it is desired to use 30 control rules as a control object, one arithmetic unit will not be sufficient, so three fuzzy inference arithmetic units will be used in parallel.

The present invention relates to a method for operating a plurality of fuzzy inference calculation devices in parallel, which is necessary in such a case.

[Conventional technology]

First, before explaining the direct prior art of the present invention, regarding the fuzzy inference arithmetic device which is the premise of the present invention,
Please explain as briefly as necessary.

Now, what about fuzzy inference operations? When performing fuzzy inference using fuzzy variables expressed by membership functions for the antecedent and consequent propositions of Rule 11, the synthesis side of inference is generally often used. That is, when a plurality of control rules are given, the ambiguous relationship R is determined in advance, and the membership function for the output is determined using the input value expressed by the membership function.

FIG. 2 is an explanatory diagram showing an example of an arithmetic device that simplifies such conventional fuzzy inference. Reference is made to FIG. 2 below.

Suppose that there is a control rule expressed by the following equation.

IF xo1=A11. x02=A21. x
03=A31゜・------=-xok=Akl
T HE N u = B (If xO1= Al
l, x02=A21. x03=A31゜・
・・・・・・・・・If xok=Akl, the manipulated variable u
is B The block (A) surrounded by a broken line in Figure 2
is the part that performs the antecedent part calculation of the above control rule. In the same figure, All, A21. A31. ...Ak
i is the membership function of the antecedent part, respectively, and Bl
.

... BN is the membership function of the consequent, respectively, xOl, x02. x, 03.・xok
are measured values.

la, lb,... 1 is an arithmetic circuit that calculates the membership function value (indicated by the symbol F), and 2 is a minimum value arithmetic circuit (indicated by the symbol) that takes the minimum value from multiple inputs and outputs it. , 3 is a circuit (indicated by *) that multiplies the input membership function B1 by a constant (multiple the output of minimum value calculation circuit 2) and outputs it, and 4 always selects and outputs the maximum value from among multiple human forces. This is a maximum value calculation circuit (indicated by the symbol ■).

Next, the control operation will be explained. The arithmetic circuit 1a calculates and outputs a membership function value from the membership function All of the antecedent part and the measured value xo1. Other arithmetic circuit 1b,
... The same applies to 1. The minimum value calculation circuit 2 receives each output from the calculation circuits 1a, lb, . . . 1k, selects the minimum value from among them, and outputs it as the effectiveness α1. The antecedent part calculation ends when this validity degree α1 is determined.

Next, we move on to the consequent operation. The constant multiplier circuit 3 multiplies the consequent membership function B1 by the validity α1 and uses the result as Bl
' and sends it to the maximum value calculation circuit 4. In the maximum value calculation circuit 4, the results obtained by similarly calculating other control rules are inputted as B2°...BN''.The maximum value calculation circuit 4 inputs these human power Bl'... BN'
The maximum value is always selected from among the waveforms and output as the output membership function B. To actually obtain the manipulated output value from the output membership function B, there is a method to obtain the manipulated output value by finding the center of gravity in the area formed by the function waveform of the membership function B and the horizontal axis. It has been adopted.

By the way, there are a finite number of fuzzy variables in the consequent part of the control rule (Bl, ... BN in Figure 2), and B1 represents NB (negative and large) and B2 represents NM (negative and medium). ,
B3 is NS (negative and small), B4 is ZE (as is), B5 is PS (positive and small), B6 is PM (positive and medium), and B7 is PB (positive and large). The fuzzy variable of the consequent part consists of these seven stages.

However, the number of control rules is not limited to seven, and a larger number may exist. However, even in that case, the fuzzy variables in the consequent part of each control rule are the above seven, that is, N
B (negative and large), NM (negative and medium), NS (negative and small), ZE (as is), PS (positive and small)
, PM (positive and medium), and PB (positive and large).

Therefore, it is necessary to check in advance which control rule corresponds to which consequent fuzzy variable and make the correspondence (
.

Returning to FIG. 2, the maximum value calculation circuit 4 always selects and outputs the maximum value from among the waveforms of a finite number of input fuzzy variables (81'...BN').

This means that if there are multiple control rules corresponding to B1', the fuzzy variable Bl' is calculated by calculating the consequent part of each control rule to find the value of the fuzzy variable Bl', and then using the maximum value calculation circuit. 4, the maximum value arithmetic circuit 4 only uses the maximum value h among the values belonging to the fuzzy variable Bl', and discards values below that value.

If this is the case, even if various values are given for the effectiveness α to be calculated between the fuzzy variable B1 and the fuzzy variable B1, even if various values are given from multiple control rules, all but the highest value α will be In the end, it is useless to perform calculations between the two.

Therefore, for fuzzy variable B1, among the various values of α given from multiple control rules for B1, only α that indicates the maximum value is stored as a grade, and only for that grade is the final difference between the fuzzy variable 81 and You can calculate it by using . The same applies to other fuzzy variables B2 and subsequent ones. That is, the highest effectiveness degree α belonging to the same fuzzy variable is selected and stored in the storage device as a grade. Hereinafter, this storage device will be referred to as a grade storage device.

In other words, the grade read from the grade storage device is used for the calculation, and the waveform of the output membership function is determined by performing the consequent calculation for each of the consequent membership functions.

FIG. 3 is an explanatory diagram of the storage contents of such a grade storage device. In the figure, it is assumed that there are seven consequent fuzzy variables from B1 to 87, and the grades obtained for each are stored as α1 to α7.

Now, with the above as background knowledge, the direct prior art of the present invention will be explained below.

FIG. 4 is a block diagram showing a conventional method for operating multiple fuzzy inference calculation devices in parallel. In the same figure, la,
lb is a control rule storage device, 2a and 2b are measured value input devices, 3a and 3b are fuzzy inference calculators, 4 is a maximum value calculator, 5 is a manipulated value calculator, and 6 is a manipulated value output device. .

This case shows a case where two fuzzy inference arithmetic units operate in parallel. That is, the fuzzy inference calculator 3a performs fuzzy inference calculations by receiving control rules (including antecedent and consequent propositions) from the storage device 1a and measurement values from the human power device 2a. Create and output a membership function for output. Similarly, the fuzzy inference calculator 3b receives control rules from the storage device 1b and measurement values from the input device 2b, performs fuzzy inference calculations, and creates and outputs an output Menharsisop function.

The maximum value calculator 4 compares the waveform of the output membership function input from the fuzzy inference calculator 3a with the waveform of the output membership function input from the fuzzy inference calculator 3b, and takes a higher value. Select the waveform part and output as the maximum value.

The manipulated value calculator 5 receives this, calculates the manipulated value, sends it to the manipulated value output device 6, and outputs it as a manipulated value.

[Problem that the invention seeks to solve]

In the conventional parallel operation method described above, each fuzzy inference calculator independently performs inference operations including both antecedent and consequent operations of a control rule, which is wasteful. There is a problem. That is, each fuzzy inference calculator must have a storage device for the consequent part nonvarship function for the consequent part calculation, which increases the cost.

An object of the present invention is to solve the problem of high cost in the conventional method of operating multiple fuzzy inference arithmetic devices in parallel.

Means for Solving Problem c] In order to solve the problem, in the present invention, the fuzzy inference calculator is divided into an antecedent part calculator that performs only antecedent part calculations and a consequent part calculator that performs only consequent part calculations. For each antecedent part calculator, only one consequent part calculator is provided in common.

[Effect]

In the consequent part calculation unit provided in common for multiple antecedent part calculation units, the grade value read from the grade storage device of each antecedent part calculation unit is made to correspond to the same consequent part fuzzy variable. The maximum value is selected by comparing them, and the consequent part of the control rule is calculated only for the maximum value.

〔Example〕

Next, the present invention will be explained in detail with reference to bacteria.

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

In the figure, 10a and 10b are antecedent part calculators, and 20 is a consequent part calculator. The antecedent part calculator 10a is
Control rule antecedent storage device 11a, measured value input device 12a
, control rule antecedent calculation circuit 13a, grade storage device 1
Similarly, the antecedent part calculator 10b includes a control rule antecedent part storage device 11b, a measured value input device 12b, a control rule antecedent part calculation circuit 13b, and a grade storage device 14b. The consequent part calculator 20 is a grade reading device 2
1, 22, a maximum value calculator 23, a consequent part calculation circuit 25, a control rule consequent part storage device 24, and a manipulated value output device 26.

Next, the operation will be explained. In the antecedent part calculator 10a,
The control rule antecedent part calculation circuit 13a receives the control rule antecedent part from the storage device 11a and the measured value from the input device 12a, and calculates the antecedent part of a plurality of control rules, and converts the consequent part into fuzzy variables. A grade value corresponding to a one-to-one relationship is determined and stored in the grade storage device 14a.

The antecedent part calculation unit 10b also performs the antecedent part calculation in exactly the same manner, and stores the obtained grade value in the grade storage device 14b. The calculations in each antecedent part calculation unit are performed in parallel.

When the antecedent part calculation is completed, the consequent part calculation in the consequent part calculation unit 20 is started. That is, the grade reading device 21 reads the grade value from the grade storage device 14a of the antecedent portion calculator 10a, and the grade reading device 22 reads the grade value from the grade storage device 14b of the antecedent portion calculator 10b. The maximum value calculator 23 is connected to the reading device 21
．． 22 calculates the maximum value among the read grade values for each corresponding consequent part fuzzy variable, and the consequent part arithmetic circuit 25
Output to.

The consequent calculation circuit 25 uses the grade value (maximum value) and the control rule consequent (membership function representing the content of the consequent fuzzy variable) given from the storage device 24 to calculate a membership function regarding the output. search, output device 26
send to The output device 26 calculates the operation value from the membership function and outputs it.

〔Effect of the invention〕

As explained above, according to the present invention, in a method for operating multiple fuzzy inference calculation devices in parallel, a consequent part calculation unit is provided in common for a plurality of antecedent part calculation units to reduce the number. However, it has the advantage of reducing costs.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is an explanatory diagram showing an example of a conventional fuzzy inference calculation device, and FIG.
FIG. 4 is a diagram illustrating the storage contents of the grade storage device, and FIG. 4 is a block diagram showing a conventional method for operating a plurality of fuzzy inference calculation devices in parallel. Description of symbols 10a, 10b...Antecedent part calculator, 20...Consequent part calculator, lla, Ilb...Control rule antecedent part storage device, 12a, 12b--Measurement value input device, 13a, 13
b... Control rule antecedent part calculation circuit, 14a, 14b...
-Grade storage device 14a, 21.22...Grade reading device, 23...Maximum value calculation unit, 24...Control rule consequent part storage device, 25...Consequent part calculation circuit, 26.
・・Manipulated value output device

Claims (1)

[Claims] 1) When the number of given control rules to be calculated is greater than the number of types of consequent fuzzy variables, and the control rules can be separated and associated with each consequent fuzzy variable, each control A grade storage device that compares the effectiveness obtained by the antecedent part calculation of a rule with those corresponding to the same consequent part fuzzy variables, and stores the maximum among them as a grade for each consequent part fuzzy variable. A plurality of antecedent part calculators that operate only in parallel and perform only the antecedent part calculations of the control rule, and a consequent part calculator provided in common for these multiple antecedent part calculators, each Compare the grade values read from the grade storage device of the antecedent part calculation unit with those corresponding to the same consequent part fuzzy variable, select the maximum value, and perform the consequent part calculation of the control rule only for the maximum value. A method for operating multiple fuzzy inference arithmetic devices in parallel.