JPH04340628A - Fuzzy inference process compile system - Google Patents

Abstract

PURPOSE:To generate an assembly source for microcomputer adapted to execution of fuzzy inference in the synchronous system or that in the asynchronous system by user's selective setting. CONSTITUTION:A compile mode for synchronous system where the assembly source for microcomputer adapted to execution of fuzzy inference in the synchronous system where input data of inference is collectively delivered at the time of starting the inference process, a compile mode for asynchronous system where the assembly source for microcomputer adapted execution of fuzzy inference in the asynchronous system where only input data of required inference is delivered any time during the inference process in accordance with each inference rule, and an execution mode selective setting part which selectively sets execution of the compile mode for synchronous system and that for asynchronous system are provided.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fuzzy inference processing compilation system, and more particularly to a fuzzy inference processing compilation system for converting fuzzy inference knowledge into assembly source for a microcomputer.

0002

[Prior Art] In order to execute fuzzy inference processing by a microcomputer, fuzzy inference rules, input/output information such as labels used in the fuzzy inference rules, and fuzzy inference inference knowledge such as member zip functions corresponding to the labels are stored in a microcomputer. Fuzzy inference processing compilers for converting assembly sources for computers are already known.

[0003] In fuzzy inference, there is a synchronous method in which inference input data is transferred to a buffer memory etc. all at once at the start of inference processing, and a synchronous method in which only necessary inference input data is transferred each time according to each inference rule during inference processing. There is an on-demand asynchronous method for delivering data to and from a buffer memory, etc., and it is preferable to use this method depending on the field of application.

For example, in the field of control, compared to the response frequency of the controlled object, the reciprocal of the sampling period Ts is 1/
If Ts is sufficiently large, there will be no large difference in control performance between synchronous and asynchronous fuzzy inferences, but as 1/Ts approaches the response frequency of the controlled object, it becomes clear whether fuzzy inference is synchronous or not. , whether it is an asynchronous method, in other words, it uses the information of the previous cycle or the information of the latest cycle being executed, resulting in a difference in control performance, and the information of the latest cycle being executed. Better control performance can be obtained with asynchronous fuzzy inference that can use

On the other hand, in various fields of recognition, the simultaneity of input information used for inference processing is important, so synchronous fuzzy inference is required.

[0006] Furthermore, the synchronous method and the asynchronous method equally exist for output data of fuzzy inference.

[0007]

[Problems to be Solved by the Invention] As mentioned above, the execution method of fuzzy inference should be used flexibly depending on the field of application of fuzzy inference, but the created fuzzy inference inference knowledge can be used as an assembly source for a microcomputer. When converting the assembly source to Compilers that create assembly sources for microcomputers suitable for execution have different contents, and there is no single compiler that can selectively do both.

The present invention has been made by focusing on the above-mentioned problems in conventional compilers for fuzzy inference processing, and is aimed at creating an assembly source for a microcomputer that is suitable for executing synchronous fuzzy inference. It is an object of the present invention to provide a fuzzy inference processing compilation system that can create assembly source for a microcomputer suitable for execution of asynchronous fuzzy inference based on user selections and settings.

[0009]

[Means for Solving the Problems] According to the present invention, in a fuzzy inference processing compilation system that converts fuzzy inference knowledge into assembly source for a microcomputer, inference processing is started on input data for inference. Synchronous compile mode that creates microcomputer assembly source suitable for execution of synchronous fuzzy inference, which is delivered in bulk at times, and only the necessary inference input data is delivered each time according to each inference rule during inference processing. an asynchronous compile mode for creating a microcomputer assembly source suitable for executing asynchronous fuzzy inference; and an execution mode selection setting for selecting and setting execution between the synchronous compile mode and the asynchronous compile mode. This is achieved by a fuzzy inference processing compilation system characterized by having a section.

0010

[Operation] According to the above configuration, the execution mode selection setting section arbitrarily sets the execution mode of compilation to either the synchronous method compilation mode or the asynchronous method compilation mode, and the synchronous method The fuzzy inference inference knowledge is converted into assembly source for a microcomputer by executing either the compile mode or the asynchronous compile mode, and the assembly source for a microcomputer and the asynchronous fuzzy inference that are suitable for execution of synchronous fuzzy inference are converted. Any assembly source for a microcomputer that is suitable for executing the inference is selectively created.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the accompanying drawings.

The fuzzy inference processing compilation system according to the present invention, as shown in FIG. Compile unit 1 for synchronous method to create, and assembly source for microcomputer suitable for execution of asynchronous fuzzy inference, which transfers only input data for inference required each time according to each inference rule during inference processing, for asynchronous method. It includes a compile section 2 and an execution mode selection and setting section 3 that selects and sets execution of either the compile section 1 for synchronous method or the compile section 2 for asynchronous method.

This fuzzy inference processing compilation system is executed by a general microcomputer or a dedicated compile machine, and is given fuzzy inference knowledge created by an appropriate editor, fuzzy inference knowledge creation tool, etc. When converting knowledge into assembly source for a microcomputer, first execute the execution mode selection setting section 3 and specify on the CRT of the execution machine whether to compile in synchronous compile mode or asynchronous compile mode. Displays a screen for selecting and setting. This mode selection is performed manually using the keyboard, mouse, dip switch, etc. When the synchronous compilation mode is selected,
The synchronous compiler 1 is executed to create a microcomputer assembly source suitable for executing synchronous fuzzy inference in which input data for inference is transferred all at once at the start of inference processing. On the other hand, when the asynchronous compile mode is selected, the asynchronous compile unit 2 is executed and is compatible with the execution of asynchronous fuzzy inference, which transfers only the necessary inference input data each time according to each inference rule during inference processing. Then, assembly source for the microcomputer is created. The flow of this mode selection and compilation execution is shown in FIG.

Next, as a concrete example, four input data in
1, in2, in3, in4, and the following five inference rules rule1, rule2, rule3, rule
4. rule5 and two output data out1, out
The processing procedure in synchronous fuzzy inference and the processing procedure in asynchronous fuzzy inference when fuzzy inference based on 2 is executed by a microcomputer as shown in FIG. 3 will be explained in order using FIGS. 4 and 5. explain.

First, the above inference rules will be described. (rule1) If in1=P and in2=Pthen
out1=P, out2=N (rule2) If in1=N and in2=Nthen
out1=N, out2=P(rule3) If in2=P and in3=Pthen
out2=N (rule4) If in3=P and in4=Pthen
out2=P (rule5) If in3=N and in4=Nthin
out2=N Next, the microcomputer shown in FIG. 3 will be explained. This microcomputer includes a CPU 10 including an input buffer and an output buffer, and input data in1, in2, i
Stores the A/D converter 11 that A/D converts n3 and in4, the D/A converter 12 that D/A converts the output data out1 and out2, system programs, assembly source files for fuzzy inference processing, etc. memory 13
It has

Next, the processing procedure in synchronous fuzzy inference will be explained with reference to FIG. First, all four input data in1, in2, in3, and in4 are read from the A/D converter 11 and stored in the input buffer (step 10). When this storage is completed, input data in1 and in2 are read from the input buffer (step 20), and inference rules rule 1 and rule 2 are sequentially executed (step 30). Next, inference rule rule
Output data obtained by executing 1 and rule 2 out
A process is performed to obtain a definite value of 1 (step 40), and this is written to the output buffer (step 50).

Next, input data in3 is read from the input buffer (step 60), and inference rule rule 3 is executed (step 70). Then, input data in4 is read from the input buffer (step 80), and inference rules rules 4 and 5 are executed (step 90). Next, a process is performed to determine the final value of the output data out2 obtained by executing the inference rules rules 1 to 5 (step 100).
), write this to the output buffer (step 110)
.

Finally, the output data out1 of the output buffer
, 2 to the D/A converter 12 (step 120).
.

Next, the processing procedure in asynchronous fuzzy inference will be explained with reference to FIG. In this case, first, input data in1 and in2 necessary for executing inference rules rule1 and rule2 are read from the A/D converter 11 and stored in the input buffer (step 10). When this store is completed, the input data in1 is transferred from the input buffer.
, in2 (step 20), and inference rule ru
Execute le1 and rule2 sequentially (step 30)
. Next, a process is performed to obtain the final value of the output data out1 obtained by executing the inference rules rule 1 and rule 2 (step 40), and this is written to the output buffer (
step 50), and the output data out1 is immediately
/Write to A converter 12. (Step 60).

Next, input data in3 necessary for executing inference rule rule 3 is read from the A/D converter 11 and stored in the input buffer (step 70). When this store is completed, the input data is input from the input buffer.
3 (step 80), inference rule rule 3
(Step 90).

Next, input data in4 necessary for executing inference rules rule 4 and rule 5 is read from the A/D converter 11 and stored in the input buffer (step 100). When this storage is completed, input data in4 is read from the input buffer (step 110), and inference rules rule 4 and rule 5 are executed sequentially (step 120). Next, inference rules rule 1 to rule 5
The final value of the output data out2 obtained by the execution of is determined (step 130), this is written to the output buffer (step 140), and the
ut2 is immediately written to the D/A converter 12. (Step 150).

[0022] In both the synchronous and asynchronous fuzzy inference processing as described above, the sampling period of input data,
For example, it is executed every sampling period Ts as shown in FIG. In synchronous fuzzy inference processing, for example, in fuzzy inference processing in a cycle of Tn, Tn
-1 input data in1, in2, i in the same period
In contrast, in asynchronous fuzzy inference processing, Tn-1 or in4 is used as the latest input data at the time of execution, depending on the execution timing of the inference rule in Tn-1 fuzzy inference processing. Input data in1, in2, in3, and in4 in the period Tn are used.

In the above embodiment, the synchronous method and the asynchronous method are respectively used. If the input data is synchronous, the output data is also synchronous, and if the input data is asynchronous, the output data is also asynchronous. , the synchronicity and asynchrony of the output data do not necessarily have to match that of the input data; for example, the input data may be synchronous and the output data may be asynchronous. In this case, the execution mode The selection setting section 3 only needs to separately have a mode selection setting function for input data and a mode selection setting function for output data.

[0024]

[Effects of the Invention] As understood from the above explanation, according to the fuzzy inference processing compilation system according to the present invention,
The execution mode selection setting section arbitrarily sets the compilation execution mode to either synchronous compilation mode or asynchronous compilation mode, and depending on this setting, either synchronous compilation mode or asynchronous compilation mode is selected. The fuzzy inference inference knowledge is converted into microcomputer assembly source by executing , and the microcomputer assembly source is adapted to execute synchronous fuzzy inference and the microcomputer assembly source is adapted to execute asynchronous fuzzy inference. Either one of these is selectively created, and as a result, it is possible to create an assembly source for a microcomputer that is suitable for executing fuzzy inference using a synchronous method, or an assembly source for a microcomputer that is suitable for executing fuzzy inference using an asynchronous method. will be performed based on selection settings by the user.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a configuration example of a fuzzy inference processing compilation system according to the present invention.

FIG. 2 is a flowchart showing the operation of the fuzzy inference processing compilation system according to the present invention.

FIG. 3 is a block diagram showing an example of a microcomputer that executes fuzzy inference processing using an assembly source created by the fuzzy inference processing compilation system according to the present invention.

FIG. 4 is a flowchart showing an example of a processing procedure in synchronous fuzzy inference.

FIG. 5 is a flowchart showing a processing procedure in asynchronous fuzzy inference.

FIG. 6 is a time chart showing the sampling period of input data and the sampling timing of input data in fuzzy inference processing.

[Explanation of symbols]

1 Compilation section for synchronous method 2 Compilation part for asynchronous method 3 Execution mode selection setting section 10 CPU 11 A/D converter 12 D/A converter 13 Memory 13

Claims (1)

[Claims] Claim 1: A fuzzy inference processing compilation system that converts fuzzy inference knowledge into assembly source for a microcomputer, a microcomputer adapted to execute synchronous fuzzy inference in which input data for inference is transferred all at once at the start of inference processing. Create assembly sources for microcomputers that are compatible with the synchronous compile mode that creates assembly source for computers, and the execution of asynchronous fuzzy inference that passes only the necessary inference input data each time according to each inference rule during inference processing. A fuzzy inference processing compile comprising: an asynchronous compile mode, and an execution mode selection setting section that selects and sets execution of either the synchronous compile mode or the asynchronous compile mode. system.