Data processor for performing a fuzzy logic weighting function and method therefor
First Claim
1. In a fuzzy inference system comprising a plurality of fuzzy rules including input labels as antecedents and output labels as consequents, a minimum-maximum computing circuit for fuzzy inference for executing minimum-maximum operations on input label grades, the minimum-maximum computing circuit comprising:
- a rule memory for storing rule-associative-bit-groups in memory areas, the memory areas being addressed by label codes, the label codes for discriminating among said input labels, each of said rule-associative-bit-groups comprising at least one coded rule, each of the at least one coded rule including a plurality of valid/invalid bits arranged in a predetermined order, each of the plurality of valid/invalid bits associated with an input label, each of the plurality of valid/invalid bits indicating an antecedent inclusion status of a corresponding input label;
an input label sorting means for sorting said input label grades and correlated label codes in order of grade magnitude and outputting sorted input label grades in order of magnitude, said input label sorting means simultaneously outputting said correlated label codes to be provided to an address input of said rule memory for reading said rule-associative-bit-groups;
a weighting factor outputting means for outputting weighting factor values in order of magnitude to be applied to at least one of said plurality of fuzzy rules;
a weighting factor-associative-bit-groups generating means for storing predetermined weighting factors and generating weighting factor-associated-bit-groups comprising an arrangement of valid/invalid bits for indicating whether each of said predetermined weighting factors stored by the weighting factor-associated-bit-groups generating means is equal or not to said weighting factor values generated by said weighting factor generating means;
a selecting means for comparing each of the input label grades output from said input label sorting means and each of the weighting factor values output from said weighting factor output means to select one of said rule-associative-bit-groups output from said rule memory and said weighting factors-associative-bit-groups output from said weighting factor-associative-bit-groups generating means according to the result of said comparison, said selecting means also selecting one of said input label grades output from said input label sorting means and selecting weighting factor values output from said weighting factor output means according to the result of said comparison; and
a minimum-maximum detecting means being arranged for each of said output labels for detecting minimum-maximum values for each of said output labels, said minimum-maximum detecting means first detecting information about minimum values among said input label grades and weighting factors selected by said selecting means, said detection of information being carried out for each of said rules included in each of said output labels according to whether a significant valid bit appeared first or last in each of said coded rules, said minimum-maximum detecting means then detecting a maximum value among said minimum values for each of said rules included in each of said output labels according to said detected information about the minimum values.
2 Assignments
0 Petitions
Accused Products
Abstract
A minimum-maximum computation circuit (5) includes a circuit (10) for detecting an order of magnitude of the input label'"'"'s grades and a circuit for executing a minimum-maximum calculation (41, 42, and 49) according to an order of the magnitude of the input label'"'"'s grades. The minimum-maximum computation circuit includes a rule memory (20) for storing rule-associative-bit-groups which include a valid/invalid bit for each of the coded rules. Each of the coded rules include an arrangement of valid/invalid bits for indicating whether each of said input labels is included or not in the antecedent of said each of rules. The minimum-maximum computing circuit further includes a weighting factor circuit (63) for providing weighting factors in their magnitude order to be applied to selected fuzzy rules.
-
Citations
16 Claims
-
1. In a fuzzy inference system comprising a plurality of fuzzy rules including input labels as antecedents and output labels as consequents, a minimum-maximum computing circuit for fuzzy inference for executing minimum-maximum operations on input label grades, the minimum-maximum computing circuit comprising:
-
a rule memory for storing rule-associative-bit-groups in memory areas, the memory areas being addressed by label codes, the label codes for discriminating among said input labels, each of said rule-associative-bit-groups comprising at least one coded rule, each of the at least one coded rule including a plurality of valid/invalid bits arranged in a predetermined order, each of the plurality of valid/invalid bits associated with an input label, each of the plurality of valid/invalid bits indicating an antecedent inclusion status of a corresponding input label; an input label sorting means for sorting said input label grades and correlated label codes in order of grade magnitude and outputting sorted input label grades in order of magnitude, said input label sorting means simultaneously outputting said correlated label codes to be provided to an address input of said rule memory for reading said rule-associative-bit-groups; a weighting factor outputting means for outputting weighting factor values in order of magnitude to be applied to at least one of said plurality of fuzzy rules; a weighting factor-associative-bit-groups generating means for storing predetermined weighting factors and generating weighting factor-associated-bit-groups comprising an arrangement of valid/invalid bits for indicating whether each of said predetermined weighting factors stored by the weighting factor-associated-bit-groups generating means is equal or not to said weighting factor values generated by said weighting factor generating means; a selecting means for comparing each of the input label grades output from said input label sorting means and each of the weighting factor values output from said weighting factor output means to select one of said rule-associative-bit-groups output from said rule memory and said weighting factors-associative-bit-groups output from said weighting factor-associative-bit-groups generating means according to the result of said comparison, said selecting means also selecting one of said input label grades output from said input label sorting means and selecting weighting factor values output from said weighting factor output means according to the result of said comparison; and a minimum-maximum detecting means being arranged for each of said output labels for detecting minimum-maximum values for each of said output labels, said minimum-maximum detecting means first detecting information about minimum values among said input label grades and weighting factors selected by said selecting means, said detection of information being carried out for each of said rules included in each of said output labels according to whether a significant valid bit appeared first or last in each of said coded rules, said minimum-maximum detecting means then detecting a maximum value among said minimum values for each of said rules included in each of said output labels according to said detected information about the minimum values. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9)
-
-
10. In a fuzzy inference system comprising a plurality of fuzzy rules including input labels as antecedents and output labels as consequents, a minimum-maximum computing circuit for fuzzy inference for executing minimum-maximum operations on input label grades, the minimum-maximum computing circuit comprising:
-
a rule memory for storing rule-associative-bit-groups in memory areas, the memory areas being addressed by label codes, the label codes for discriminating among said input labels, each of said rule-associative-bit-groups comprising at least one coded rule, each of the at least one coded rule including a plurality of valid/invalid bits arranged in a predetermined order, each of the plurality of valid/invalid bits associated with an input label, each of the plurality of valid/invalid bits indicating an antecedent inclusion status of a corresponding input label; a weighting factors-associative-bit-groups memory for storing weighting factors-associative-bit-groups in memory areas accessed by weighting factors codes for discriminating weighting factors from each other, said weighting factors being applied to at least one of said fuzzy rules; an input labels and weighting factors sorting means for sorting said input label grades and said weighting factors together with correlated input label codes and correlated weighting factor codes according to the order of magnitude of said input label grades and weighting factors, and outputting said sorted input label grades and weighting factors in magnitude order, said input label sorting means simultaneously outputting said correlated input label codes and said correlated weighting factor codes to be provided to an address input terminal of said rule memory and said weighting factors-associative-bit-groups memory to cause reading out of said rule-associative-bit-groups and weighting factors-associative-bit-groups, each of the bits of said rule-associative-bit-groups comprising each of said coded rules when read out successively; and a minimum-maximum detecting means being arranged for each of said output labels for detecting minimum-maximum values for each of said output labels, said minimum-maximum detecting means first detecting information about minimum values among said input label grades and weighting factors read out from said rule memory and said weighting factors-associative-bit-groups memory, said minimum-maximum detecting means detecting information for each of said rules included in each of said output labels according to whether a significant valid bit appeared first or last in each of said coded rules, said minimum-maximum detecting means then detecting a maximum value among said minimum values included in each of said output labels according to said detected information about the minimum values. - View Dependent Claims (11, 12, 13, 14, 15, 16)
-
Specification