Fuzzy backward reasoning system and expert system utilizing the same

• US 5,263,123 A
• Filed: 03/11/1991
• Issued: 11/16/1993
• Est. Priority Date: 09/10/1990
• Status: Expired due to Fees

• Current Assignee: Hiachi Engineering Company Limited
• Original Assignee: Hitachi Engineering Co.,Ltd (Hitachi, Ltd.)
• Inventors: Hayashi, Yoichi
• Primary Examiner(s): MacDonald, Allen R.
• Application Number:
  US07/666,923
• Time in Patent Office: 981 Days
• Field of Search: 395/3, 395/61, 395/900, 395/51, 395/52
• US Class Current: 706/45
• CPC Class Codes: G05B 13/0295 using fuzzy logic and expert systems
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: G06N 7/023 Learning or tuning the parameters of a fuzzy system
• CPC Class Codes: Y10S 706/90 Fuzzy logic

Abstract

A system for controlling a target apparatus includes a supplying section, a reasoning section, and a control section. The supplying section supplies n states of the target apparatus (n≧2 and an integer). The reasoning section receives the n sensed states as n results from the supplying section and performs fuzzy backward reasoning for data representing the n results and causality data to generate certainties of m causes (m≧2 and an integer). The control section controls the target apparatus in accordance with the certainties of the causes.

Claims

1. An expert system for controlling a target apparatus comprising:

supplying means for supplying n states of the target apparatus (n is an integer and larger than or equal to

     2);

reasoning means for receiving the n states as n results from said supplying means and performing fuzzy backward reasoning for a data representing the n results and causality data to generate certainties of m causes (m is an integer and larger than or equal to

     2),

wherein each of the certainties is represented in a cause matrix table having m elements, each of which is represented by an interval data, and the causality data is represented by a causality matrix table having m×

n elements, each of which indicates a membership grade of a corresponding cause to a corresponding result and is represented by an interval data, and

wherein in the fuzzy backward reasoning said reasoning means converts the n result data into a result matrix table having n elements, each of which is represented by an interval data, performs ε

- and ε

-compositions for the result and causality matrix tables to generate first and second matrix tables, generates from the first matrix table a third matrix table having m elements, each of which represent a least upper bound value of a corresponding row of the first matrix table, generates from the first and second matrix tables a check matrix table having m×

n elements, each of which represents a possibility of influence to the certainty of a corresponding cause, and generates the cause matrix table from the first, third and check matrix tables, the cause matrix table element interval data representing a certainty of a corresponding cause;

control means for controlling the target apparatus based on the certainties of the elements of the cause matrix table.

2. A system according to claim 1, wherein said reasoning means includes:

sampling means for sampling the n states to generate the n results; and

converting means for converting the n results into the result matrix table using a plurality of conversion tables.

3. A system according to claim 1, wherein said reasoning means includes:

sampling means for sampling the n states to generate the n results;

first converting means for converting the n results into a LTV matrix table using a plurality of first conversion tables; and

second converting means for converting the n LTV table into the result matrix table using a plurality of second conversion tables and a plurality of predetermined values.

4. A system according to claim 1, wherein said reasoning means further comprises:

storing means for storing a plurality of membership function tables;

converting means for reading out the membership function tables from said storing means and converting the causality data into the causality matrix table by applying an associated one of the read out membership function tables with an associated predetermined value to each element of the causality data.

5. A system according to claim 1, wherein said reasoning means requires a theoretical operation time proportional to a value of m×

n.

6. A system according to claim 1, wherein said control means controls the target apparatus based on a cause corresponding to an element having a maximum certainty of the cause matrix table.

7. A system according to claim 1, wherein said control means controls the target apparatus based on causes corresponding to elements of the cause matrix table having certainties higher than a predetermined certainty.

8. A system according to claim 1, wherein said reasoning means comprises:

memory means for storing the causality matrix table;

first generating means for reading out the causality matrix table from said memory means and generating the first and second matrix tables from the causality table and the result matrix table from the n result data;

second generating means for generating the third matrix table having m elements, each of which represent a least upper bound value of a corresponding row of the first matrix table;

third generating means for generating from the first and second matrix tables the check matrix table having m×

n elements, each of which has a flag set when representing a possibility of influence to a certainty of a corresponding cause;

modifying means for modifying the check matrix table to selectively reset each flag in accordance with corresponding elements of the first and third matrix tables; and

fourth generating means for generating from elements having the flags set of the check matrix table and the first and third matrix tables cause the matrix tables.

9. A system according to claim 8, wherein said reasoning means requires a theoretical operation time proportional to a value of m×

n.