System for synthesizing new fuzzy rule sets from existing fuzzy rule sets

• US 5,179,634 A
• Filed: 05/28/1992
• Issued: 01/12/1993
• Est. Priority Date: 04/14/1989
• Status: Expired due to Term

• Current Assignee: U Krowel LLC (Intellectual Ventures LLC)
• Original Assignee: Omron Corporation
• Inventors: Matsunaga, Nobutomo, Shoji, Kazuaki
• Primary Examiner(s): MacDonald, Allen R.
• Application Number:
  US07/888,435
• Time in Patent Office: 229 Days
• Field of Search: 395/61, 395/900, 395/75, 395/51
• US Class Current: 706/59
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: G06N 7/026 Development tools for entering the parameters of a fuzzy system
• CPC Class Codes: Y10S 706/90 Fuzzy logic

Abstract

A fuzzy rule generator for generating fuzzy rules, comprising: a selection unit for selecting two or more sets of rules from a plurality of basic sets of rules which may be stored in memory; and a control unit for generating a new set of rules by carrying out a synthetic arithmetic operation based on a predetermined set of rules for synthesis. This process of synthesis may simply consist of addition or subtraction. Thereby, a new set of rules can be easily generated by carrying out a simple arithmetic operation for each application. Therefore, generation and modification of rules are simplified and speeded up, with additional benefits arising from an efficient utilization of existing sets of rules. Further, by imposing a constraint on the newly synthesized set of rules manually, automatically or by a combination of both according to a predefined algorithm, it is possible not only to ensure the stability of the system but also to obtain desired control characteristics. If desired, the newly created sets of rules may be added to the library of the basic sets of fuzzy rules.

Claims

1. A fuzzy rule generator for generating new fuzzy rules based upon a plurality of existing basic sets of fuzzy rules, comprising:

a memory device for storing said plurality of basic sets of fuzzy rules;

selection means coupled to said memory device for accessing said memory device and selecting therefrom at least two sets of existing basic sets of fuzzy rules from said stored plurality of existing basic sets of fuzzy rules; and

a rule synthesizer connected to an output of said selection means which produces a new set of fuzzy rules by combining together the selected at least two basic sets of fuzzy rules in accordance with a predetermined set of rules for combining said selected at least two sets of fuzzy rules to form said new set of fuzzy rules.

2. A fuzzy rule generator according to claim 1 further comprising a fuzzy logic inference device which operates an applied input data using said new set of fuzzy rules.

3. A fuzzy rule generator according to claim 1, wherein said combining of said selected at least two sets of rules consists of adding together said selected at least two sets of rules.

4. A fuzzy rule generator according to claim 1, wherein said combining of said selected at least two sets of rules consists of subtracting from each other said selected at least two sets of rules.

5. A fuzzy rule generator according to claim 1, wherein said rule generating means further comprises:

means for determining whether said selected at least two basic sets of rules include conflicting rules; and

means for modifying one of said selected basic sets of rules according to a prescribed rule when said selected basic sets of rules include conflicting rules.

6. A fuzzy rule generator according to claim 1, further comprising a rule modifier connected to said rule synthesizer for imposing a predetermined constraint relating to a control object on said new set of fuzzy rules to ensure a stability of said control object.

7. A fuzzy rule generator according to claim 6, wherein said rule modifier comprises manual input means for modifying a part of said new set of fuzzy rules.

8. A fuzzy rule generator according to claim 7, wherein said rule modifier further includes means for examining a Liapunov function defined for said control object, and means connected to said means for examining for modifying said new set of fuzzy rules so as to ensure stability of said control object.

9. A fuzzy rule generator according to claim 7, wherein said rule modifier further includes means for examining a Liapunov function defined for said control object, and manual input means connected to said means for examining for allowing said new set of fuzzy rules to be manually modified so as to obtain a desired control property within a range of fuzzy rules ensuring a stability of said control object.

10. A method for generating new fuzzy rules based upon a plurality of existing basic sets of fuzzy rules, comprising:

storing said plurality of existing basic sets of fuzzy rules in a memory device;

selecting from said memory device at least two sets of said existing basic sets of fuzzy rules from said plurality of stored existing basic sets of fuzzy rules; and

synthesizing a new set of fuzzy rules by combining the selected at least two sets of fuzzy rules in accordance with a predetermined set of combination rules to form the new set of fuzzy rules.

11. The method of claim 10 further comprising the step of using said new set of fuzzy rules in a fuzzy logic inference device.

12. The method of claim 10, wherein said step of synthesizing a new set of fuzzy rules comprises adding together said selected at least two sets of fuzzy rules.

13. The method of claim 10, wherein said step of synthesizing a new set of fuzzy rules comprises subtracting from each other said selected at least two sets of fuzzy rules.

14. The method of claim 10, further comprising the steps of:

determining whether said selected at least two basic sets of rules include conflicting rules; and

modifying one of said selected basic sets of rules according to a prescribed rule when said selected basic sets of rules include conflicting rules.

15. The method of claim 10, further comprising the step of imposing a predetermined constraint on said generated new set of fuzzy rules to ensure a stability of a control object.

16. The method of claim 15, wherein said predetermined constraint is used to modify a part of said generated new set of rules.

17. The method of claim 15, wherein said step of imposing a constraint comprises the steps of:

examining a Liapunov function defined for its control objects; and

modifying said generated new set of fuzzy rules in accordance with said examined Liapunov function so as to ensure stability to said control object.

18. The method of claim 15, herein said steps of imposing a constraint comprises the steps of:

examining a Liapunov function defined for its control object; and

manually modifying said generated new set of fuzzy rules in accordance with said examined Liapunov function used to obtain a desired control property with a range of fuzzy rules for ensuring a stability to said control object.