JP3055902B2 - Monitoring device for fuzzy inference device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a monitor device for monitoring a fuzzy inference device used in a fuzzy computer or a fuzzy controller.

<Prior Art> As a conventional monitoring device of a fuzzy inference device, there is a device that monitors a membership function obtained as a result of fuzzy inference. According to this type of monitor, the inference result is displayed only to the observer, which is still inconvenient for the rule creator and the coordinator of the fuzzy inference device.

<Problem to be Solved by the Invention> An object of the present invention is to provide a monitor device of a fuzzy inference device that improves the function of the monitor device and is convenient for a rule creator and a coordinator of the fuzzy inference device.

<Means for Solving the Problems> In order to achieve the above-mentioned object, a fuzzy inference apparatus for monitoring a fuzzy inference apparatus according to the present invention employs a fuzzy inference apparatus that repeatedly executes fuzzy inference in a time series using a plurality of fuzzy rules. In the monitoring device of the fuzzy inference device to be monitored, a rule output unit that detects a rule that has the greatest influence on the inference output of the fuzzy inference device and outputs a detection result (in the embodiment, the “dominant rule detection unit 41 ") And a value indicating the degree of difference between the target value and the current value relating to the state of the controlled object controlled by the manipulated variable based on the inference output of the fuzzy inference device (in the embodiment, the" deviation value S "))
An evaluation signal output unit that outputs an evaluation signal of the fuzzy inference output (corresponding to the “inference result evaluation unit 42” in the embodiment); a rule output by the rule output unit; And an output device for associating the output evaluation signals with each other and outputting them.

Further, the output device may be a display device (Claim 2), a printer (Claim 3), or another device (for example, an audio output device).
Further, the fuzzy inference apparatus includes, for example, a rule processing unit that processes a plurality of fuzzy rules, and a unit that receives an output of the rule processing unit and performs a predetermined inference process. It is possible to use one that receives each output of the processing unit and the output of the fuzzy inference device, detects and outputs the rule that has the greatest effect on the output of the fuzzy inference device (claim 4).

<Operation> A predetermined input is given to the fuzzy inference device, and fuzzy inference is executed based on a plurality of fuzzy rules. At this time, the first output unit (corresponding to the dominant rule detection unit 41 in the embodiment) detects the rule that has the most influence in determining the inference result, and supplies it to the output device. The output device outputs the detection result in a form that a human can perceive. Then, it is possible to know which rule influences the inference result, and for example, an adjustment process is performed based on the rule.

A specific example of a method for detecting the rule will be described below.
First, each output of the rule processing unit and the output of the fuzzy inference device are input to the rule output means. Then, the rule output means detects which rule has the greatest effect on the output of the fuzzy inference device. Then, the detected rule is output to the output device.

In the case of claim 1, the evaluation signal output means is provided, so that together with the above operation, the target value and the current value relating to the state of the controlled object controlled by the manipulated variable based on the inference output of the fuzzy inference device. A value indicating the degree of difference from the value is input to the evaluation signal output means. Then, the evaluation signal output means judges whether the inference result is good or not and gives the judgment result to the output device. The output device associates the detection result of the rule output means with the pass / fail of the inference result and outputs the result. Then, if the inference result is good, it is possible to know that the fuzzy rule output via the output device is a good rule, and if the inference result is bad, the output fuzzy rule is a bad rule. As a result, the adjustment process for the inference device is more easily performed.

<Embodiment> FIG. 1 is a block diagram of an electric circuit of a fuzzy inference device and its monitor device. R1, R2, and R3 input a state signal of a control object generated at a terminal I, and set a rule. Inference according to, and the result is deleted in the head (head truncation)
Is a rule processing unit that outputs in the form of a given membership function. Reference numeral 1 denotes a correspondent max circuit (CMAX circuit) that receives the outputs of the rule processing units R1 to R3 and outputs the maximum value for each corresponding line. 2 is C · MAX
This is a defuzzifier that receives the output of the circuit 1 and performs defuzzification. The defuzzifier 2 outputs a result of defuzzification, that is, a final value e *. The method of defuzzification is generally to find the center of gravity of the input membership function. However, the method is not limited to the center of gravity, and there are various other methods such as finding an input value having a peak or an area equal to the left and right sides. In this embodiment, the rule processing units R1 to R3, C.MAX
A circuit consisting of the circuit 1 and the defuzzifier 2 constitutes a fuzzy inference device.

Reference numeral 3 denotes a monitor device including a monitor circuit device 4, a display 5, and a recorder 6. The monitor circuit device 4 inputs the outputs of the output membership functions B1 to B3 defuzzifiers 2 of the rule processing units R1 to R3, that is, the fixed value e * and the deviation value S, and outputs the output as an example of a display device as described later. The data is provided to a certain display 5 to be displayed, and is provided to the recorder 6 to record data. The deviation value S is the absolute value of the difference between the target value of the state of the controlled object and the state signal of the controlled object generated at the terminal I.

FIG. 2 is a block diagram of the monitor circuit device 4. The monitor circuit device 4 includes a dominant rule detection unit 41, an inference result evaluation unit 42, and an output interface 43. The dominant rule detection unit 41 determines the final value e *
And the membership functions B1 to B3 are input and the final value e *
Of the rule processing units R1 to R3, which has the greatest effect on, and outputs the corresponding rule number, that is, the dominant rule number. The inference result evaluation unit 42
The deviation value S is input, the absolute value of the deviation value S is generated by an absolute value circuit (not shown), the value is integrated by a resistor 421 and a capacitor 422, the level is discriminated by a binarization circuit 423, and an evaluation signal is output. Give q. When the evaluation signal q is at the high level H, the dominant rule has an adverse effect on the final value e * of the inference result. When the notation signal q is at the low level L,
Dominant rules have a positive effect on inference results.
The output interface 43 inputs and outputs the determined value e *, the evaluation signal q, and the dominant rule number n.

FIG. 3 is a block diagram of the dominant rule detection unit 41. The determined value e * from the defuzzifier 2 is input to the dominant rule detection unit 41, and is converted into a digital amount by the A / D converter 411. This value is latched by the latch circuit 412. The output eA of the latch circuit 412 is selected from the signal lines constituting the membership functions B1, B2, B3, which are the inputs of the analog multiplexers 413, 414, 415.
One signal line is specified for each. Each analog multiplexer 413, 414, 415 selects the signal line of the number designated by eA from among the signal lines B11 to B1N, B21 to B2N, B31 to B3N constituting the membership functions B1, B2, B3, and b1 respectively. , b2, b3. On b1, b2, b3, a voltage signal representing the membership value at the value e * of the membership functions B1, B2, B3 is placed. The analog signals b1, b2, b3 representing the membership values are respectively
Input to the A / D converters 416, 417, 418, and digital signals d1, d
It is converted to 2, d3 and input to the maximum position detection unit 419. The maximum position detection unit 419 generates a dominant rule number n by the following calculation and outputs it to the output interface 43.

FIG. 4 shows a display example on the display 5 of FIG. The determined value e *, the evaluation signal q, and the dominant rule number n are given to the display 5 of FIG. 1 through the output interface 43 of FIG. As a result, the display 5
Displays data as shown in FIG. The horizontal axis is time
t, the vertical axis shows the graph indicating the definite value e * below. In the upper part, the pass / fail of the dominant rule is displayed at the position of the dominant rule with respect to rules 1 to 3. For example, the dominant rule at time t0 is rule 1 of the rule processing unit R1, and the rule is a fixed value e
* Has been shown to have a positive effect. Also,
The dominant rule at time tn is rule 3,
It is shown that the rule has a bad influence on the inference result e *.

<Effects> As described above, according to the monitor device of the fuzzy inference device according to the present invention, the fuzzy rule that has the most influence on the inference result obtained by executing the fuzzy inference on a predetermined input is output. Therefore, rule creation and fuzzy logic device adjustment (debugging can be performed easily) based on the rules.

Moreover, the evaluation signal output means is provided, and when outputting to the output device, the quality of the inference result is output together with the fuzzy rule, so in addition to which rule has the most influence on the inference result, The quality of the inference result is also output, so it can be determined whether the rule that has the most influence on the inference result has a good effect or a bad effect, and adjustment of the fuzzy inference device is easier and faster. Can be done.

[Brief description of the drawings]

1 is a block diagram of an electric circuit of a fuzzy inference device and its monitor device, FIG. 2 is a block diagram of a monitor circuit device shown in FIG. 1, and FIG. 3 is a block diagram of a dominant rule detecting unit shown in FIG. FIG. 4 is a block diagram showing an example of display by the display shown in FIG. 3: Monitor device 5: Display 41: Dominant rule detection unit (rule output means) 42: Inference result evaluation unit (evaluation signal output means)

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G06F 9/44 554 JICST file

Claims (4)

(57) [Claims] 1. A fuzzy inference apparatus for monitoring a fuzzy inference apparatus that repeatedly executes fuzzy inference in a time series using a plurality of fuzzy rules, wherein the fuzzy inference apparatus has the greatest influence on an inference output of the fuzzy inference apparatus. A rule output means for detecting a rule and outputting the detection result; anda degree of difference between a target value and a current value relating to a state of a controlled object controlled by an operation amount based on an inference output of the fuzzy inference apparatus. An evaluation signal output unit that outputs an evaluation signal of the fuzzy inference output upon receiving an input of a value indicating the rule, and a rule output by the rule output unit, and an evaluation signal output by the evaluation signal output unit. A monitoring device for a fuzzy inference device, comprising: an output device for outputting both in association with each other. 2. The monitoring device according to claim 1, wherein said output device is a display device. 3. A monitor device for a fuzzy inference device according to claim 1, wherein said output device is a printer. 4. The fuzzy inference apparatus comprises: a rule processing unit that processes a plurality of fuzzy rules; and a unit that receives an output from the rule processing unit and performs a predetermined inference process. The apparatus according to claim 1, wherein each of the outputs of the rule processing unit and the output of the fuzzy inference device are received, and a rule that has the greatest influence on the output of the fuzzy inference device is detected and output. 1
The monitoring device of the fuzzy inference device according to the paragraph.