Compound type expert system

• US 5,303,331 A
• Filed: 03/19/1991
• Issued: 04/12/1994
• Est. Priority Date: 03/27/1990
• Status: Expired due to Term

• Current Assignee: Ricoh Company Limited
• Original Assignee: Ricoh Company Limited
• Inventors: Namba, Yoichi
• Primary Examiner(s): Fleming, Michael R.
• Assistant Examiner(s): Downs, Robert W.
• Application Number:
  US07/671,484
• Time in Patent Office: 1,120 Days
• Field of Search: 395/61, 395/900, 395/906, 395/11, 395/51
• US Class Current: 706/10
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: Y10S 706/90 Fuzzy logic

Abstract

A compound type expert system comprising a first fuzzy expert system and a second expert system connected to the first expert system through a calculation control system. The first expert system executes fuzzy rule-based inference according to a first rule set that can describe ambiguity of language information and outputs a definite value. The second expert system executes rule-based inference according to a second rule set that can describe only information without ambiguity of language. The calculation control system is arranged between the first and second expert systems so as to activate the second expert system in response to the definite value output from the first expert system.

Claims

1. A compound type expert system comprising:

a first expert system for executing fuzzy rule-based inference according to a first rule set which permits describing information with ambiguity of language, and for outputting a definite value;

a second expert system for executing rule-based inference according to a second rule set which describes only information without ambiguity of language; and

a calculation control system arranged between said first expert system and said second expert system for calculating a plurality of control values on the basis of said outputted definite value and for activating said second expert system a number of times related to the number of said control values.

2. A compound type expert system according to claim 1, wherein said first expert system comprises a plurality of fuzzy inference engines, a MAX circuit for compounding a plurality of inference results outputted from said engines thereby to generate a fuzzy inference conclusion, and an antifuzzy circuit for calculating said definite value from said generated fuzzy inference conclusion.

3. A compound type expert system according to claim 2, wherein a plurality of said control values comprise a first control value corresponding to said definite value calculated from said fuzzy inference conclusion by said antifuzzy circuit, and a second control value and a third control value both calculated by using a maximum value and a minimum value of said fuzzy inference conclusion.

4. A compound type expert system according to claim 1, wherein said second expert system comprises a rule-base memory which registers rules describing information in a form without ambiguity of language information, an inference engine for executing rule-based inference on the basis of said rules registered in said rule-base memory, and a working memory.

5. A compound type expert system according to claim 1, wherein said calculation control system comprises a calculation control unit for creating a plurality of control values on the basis of said outputted definite value and activating said second expert system in accordance with the number of said control values, and a stack memory for registering data necessary for creating said control values and obtained during the process of creating said control values.

6. A compound type expert system according to claim 1, wherein said calculation control system and said second expert system are arranged as one unit combined together.

7. A compound type expert system comprising:

a first expert system for executing fuzzy rule-based inference according to a first rule set which permits describing information with ambiguity of language, and for outputting a definite value;

a second expert system for executing rule-based inference according to a second rule set which describes only information without ambiguity of language; and

a calculation control system arranged between said first expert system and said second expert system for calculating a plurality of control values on the basis of said outputted definite value and for activating said second expert system according to the number to said control values,

said first expert system comprising a plurality of fuzzy inference engines, a MAX circuit for compounding a plurality of inference results outputted from said engines thereby to generate a fuzzy inference conclusion, and an antifuzzy circuit for calculating said definite value from said generated fuzzy inference conclusion,

said second expert system comprising a rule-base memory which registers rules describing information in a form without ambiguity of language information, an inference engine for executing rule-based inference on the basis of said rules registered in said rule-base memory, and a working memory,

said calculation control system comprising a calculation control unit for creating a plurality of said control values on the basis of said outputted definite value and activating said second expert system according to the number of said control values, and a stack memory for registering data necessary for creating said control values and obtained during the process of creating said control values.

8. A compound type expert system according to claim 7, wherein said calculation control system and said second expert system are arranged as one unit combined together.

9. A compound type expert system operation comprising the steps of:

executing, in a first expert system, fuzzy rule-based inference according to a first rule set which permits describing information with ambiguity of language, and outputting a definite value;

calculating, in a calculation control system arranged between said first expert system and a second expert system, a plurality of control values on the basis of said outputted definite value, and activating said second expert system a number of times related to the number of said control values; and

executing, in said second expert system, rule-based inference said number of times according to a second rule set which describes only information without ambiguity of language.