Method and apparatus for setting membership functions, and method and apparatus for analyzing same

• US 5,566,274 A
• Filed: 12/17/1991
• Issued: 10/15/1996
• Est. Priority Date: 04/28/1989
• Status: Expired due to Term

• Current Assignee: U Krowel LLC (Intellectual Ventures LLC)
• Original Assignee: Omron Corporation
• Inventors: Urasaki, Kazuaki, Ishida, Tsutomu
• Primary Examiner(s): Davis, George B.
• Application Number:
  US07/768,939
• Time in Patent Office: 1,764 Days
• Field of Search: 395/900, 395/61, 395/3, 395/51
• US Class Current: 706/59
• CPC Class Codes: G05B 13/0275 using fuzzy logic only
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: G06N 7/04 Physical realisation

Abstract

A check is made to determine whether there is identity among set membership functions, different numbers are assigned to mutually different membership functions, and identical numbers are assigned to membership functions which are mutually identical. These membership functions are stored in memory in correlation with their numbers, labels and the names of input or output variables.

Claims

1. An apparatus for setting membership functions, which comprises:

a setting means for setting membership functions and corresponding identification codes for input and output variables, each of said membership functions having a predefined shape;

means for checking whether there is identity in shape among membership functions set for each input and output variable and for each identification code;

means for assigning different identification numbers to set membership functions which are different in shape from one another and identical identification numbers to set membership functions processing an identity in shape;

a first memory device for storing data, which represents a set membership function, in corresponding with the identification number assigned thereto; and

a second memory device for storing in a correlated manner names of the input or output variables, the identification numbers assigned to the differently shaped membership functions associated with respective input or output variables, and the identification codes of the membership functions which correspond to the identification numbers.

2. An apparatus for setting membership functions according to claim 1, wherein said identification codes of the membership functions are items of linguistic information representing characterizing features of the membership functions.

3. An apparatus for setting membership functions according to claim 1, further comprising means for forming membership functions of a standard shape for each of the input and output variables in dependence upon number of types of membership functions required to express positive variables and the number of types of membership functions required to express negative variables.

4. A method of setting membership functions, which comprises:

setting membership functions and corresponding identification codes for input and output variables, each of said membership functions having a predefined shape;

checking whether there is identity in shape among membership functions set for each input and output variable and for each identification code;

assigning different identification numbers to set membership functions which are different in shape from one another and identical identification numbers to membership functions possessing an identity in shape;

storing, in first memory means, data representing set membership functions in correspondence with the identification number assigned thereto; and

storing, in second memory means, and in a correlated manner, names of the input or output variables, the identification numbers assigned to the differently shaped membership functions associated with respective input or output variables, and the identification codes of the membership functions with correspond to the identification numbers.

5. An apparatus for analyzing membership functions comprising:

a setting unit for setting a plurality of membership functions;

means for calculating a degree of resemblance between two set membership functions by correlating the two set membership functions;

means for comparing the calculated degree of resemblance with a predetermined reference value and outputting an indication of at least those membership functions judged to resemble each other; and

means for deleting from said set membership functions one membership function of the set membership functions which is judged to resemble another membership function of the set, and leaving the other set membership functions intact thereby reducing a capacity of a membership-function memory.

6. An apparatus for analyzing membership functions according to claim 5,

wherein said means for calculating a degree of resemblance calculates a ratio of a maximum value of a cross-correlation function of the two membership functions to an autocorrelation value of one of the two membership functions, as well as a value of a variable of a cross-correlation function which gives the maximum value of said cross-correlation function;

and

said comparing and outputting means judges that the two membership functions resemble each other when said ratio is greater than a first reference value and the value of said variable is less than a second reference value.

7. A method of analyzing membership functions, which comprises:

setting a plurality of membership functions;

calculating a degree of resemblance between two set membership functions by correlating the two set membership functions;

comparing the calculated degree of resemblance with a predetermined reference value and outputting an indication of at least those membership functions judged to resemble each other; and

deleting one membership function from the set membership functions which is judged to resemble another membership function of the set, and leaving intact the other membership functions of the set thereby reducing a capacity of a membership-function memory.

8. A method of analyzing membership functions according to claim 7,

wherein a ratio of a maximum value of a cross-correlation functions of two membership function to an autocorrelation value of one membership function, and a value of a variable of a cross-correlation function which gives the maximum value of said cross-correlation function, are calculated;

and

the two membership functions are judged to resemble each other when said ratio is greater than a first reference value and the value of said variable is less than a second reference value.