Fuzzy logic controller having high processing speed

Claim 2

A fuzzy logic controller, said fuzzy logic controller receiving n E input variables and outputting n LW output variables, comprising: fuzzification circuit means having a memory for storing membership functions of the input variables, said fuzzification circuit means generating u n E combinations of first numbers for linguistic values of input variables addressed by externally supplied distinct values of the n E input variables and values of membership functions of the linguistic values of the input variables addressed by the externally supplied distinct values of the n E input variables, whereby a maximum of u membership functions respectively overlap and, as a result thereof, a maximum of u linguistic values are simultaneously addressed by a distinct value of the input variables; rule decoder means having a linguistic protocol and for forming, in conformity with the linguistic protocol stored therein, second numbers of linguistic values of the output variables from the u n E combinations of the first numbers, the linguistic protocol being IF/THEN rules and the u n E combinations of the first numbers being an IF part of said rules and the second members being a THEN part of said rules; rule evaluation circuit means for generating a weighting signal for each of n LW linguistic values of the output variables from the second numbers of linguistic values of the output variables and from the values of the membership functions of respective linguistic values of the output variables by means of at least one circuit for the formation of a nondistinct operation; inference circuit means having a memory and at least one further circuit for the formation of a nondistinct operation, and said inference circuit means forming signals for nondistinct unification sets at respective locations from the weighting signals for all locations of an area of distinct values of the output variables; a defuzzification circuit means having at least one evaluation circuit, said defuzzification circuit means generating distinct, externally available values of the output variables from the signals for the nondistinct unification sets of all locations of the area of distinct values of the output variables; and sequencer means for effecting that, after the u n E combinations of first numbers, the weighting signals for the n LW linguistic values of the output variables are calculated once for every set of distinct values of the n E input variables, said weighting signals being used in the inference circuit for all locations of the area of distinct values of the output variables.

Claim 3

The fuzzy logic controller according to claim 1, wherein selection signals for definition of the input variables affected by a respective rule are additionally formed in the rule decoder means; and wherein said selection signals are supplied to the rule evaluation circuit means.

Claim 4

The fuzzy logic controller according to claim 1, wherein the sequencer means also effects that nondistinct unification sets for already calculated locations of the area of distinct values of the output variables are simultaneously further-processed while a respective signal for the nondistinct unification set for a location of the area of distinct values of the output variables is calculated in the inference circuit means.

Claim 5

The fuzzy logic controller according to claim 1, wherein the rule evaluation circuit means has circuit means for formation of a limited sum and maximum circuit means that serve as first and second circuits, respectively, for the formation of a nondistinct operation; and wherein, alternatively, one of the first and second circuits is used for the formation of the weighting signals dependent on an externally supplied rule evaluation select signal.

Claim 6

The fuzzy logic controller according to claim 1, wherein the inference circuit means has circuit means for the formation of a limited sum and a maximum circuit means that serve as first and second, further circuits for the formation of a nondistinct operation; and wherein, alternatively, one of the first and second circuits is used for the formation of the signals for the nondistinct unification set at respective locations dependent on an externally suppliable inference select signal.

Claim 7

The fuzzy logic controller according to claim 1, wherein the defuzzification circuit means has a circuit for calculation of a center of gravity and a circuit for calculation of a mean of maxima that serve as first and second evaluation circuits; and wherein, alternatively, one of the first and second evaluation circuits is used for the formation of the distinct values of the output variables dependent on an externally suppliable defuzzification select signal.

Claim 8

The fuzzy logic controller according to claim 1, wherein the rule decoder means is programmable logic.

Claim 9

The fuzzy logic controller according to claim 1, wherein at least one of the memory for the input variables and the memory for the output variables is a random-access memory.

Claim 10

The fuzzy logic controller according to claim 1, wherein the fuzzy logic controller together with its memory for the input variables and its memory for the output variables is a component part of a conventional microcontroller.

Claim 11

A fuzzy logic controller, said fuzzy logic controller receiving n E input variables and outputting n LW output variables, comprising: fuzzification circuit means having a memory for storing membership functions of the input variables, said fuzzification circuit means generating u n E combinations of first numbers for linguistic values of input variables addressed by externally supplied distinct values of the n E input variables and values of membership functions of the linguistic values of the input variables addressed by the externally supplied distinct values of the n E input variables, whereby a maximum of u membership functions respectively overlap and, as a result thereof, a maximum of u linguistic values are simultaneously addressed by a distinct value of the input variables; rule decoder means having a linquistic protocol and for forming, in conformity with the linguistic protocol stored therein, second numbers of linguistic values of the output variables from the u n E combinations of the first numbers, the linguistic protocol being IF/THEN rules and the u n E combinations of the first numbers being an IF part of said rules and the second members being a THEN part of said rules, said rule decoder means having means for forming selection signals for definition of the input variables affected by a respective rule; rule evaluation circuit means for generating a weighting signal for each of n LW linguistic values of the output variables from the second numbers of linguistic values of the output variables and from the values of the membership functions of respective linguistic values of the output variables by means of at least one circuit for the formation of a nondistinct operation, said selection signals being supplied to the rule evaluation circuit means; inference circuit means having a memory and at least one further circuit for the formation of a nondistinct operation, and said inference circuit means forming signals for nondistinct unification sets at respective locations from the weighting signals for all locations of an area of distinct values of the output variables; a defuzzification circuit means having at least one evaluation circuit, said defuzzification circuit means generating distinct, externally available values of the output variables from the signals for the nondistinct unification sets of all locations of the area of distinct values of the output variables; and sequencer means for effecting that, after the u n E combinations of first numbers, the weighting signals for the n LW linguistic values of the output variables are calculated once for every set of distinct values of the n E input variables, said weighting signals being used in the inference circuit for all locations of the area of distinct values of the output variables, the sequencer means also effecting that nondistinct unification sets for already calculated locations of the area of distinct values of the output variables are simultaneously further-processed while a respective signal for the nondistinct unification set for a location of the area of distinct values of the output variables is calculated in the inference circuit means.

Claim 12

A fuzzy logic controller, said fuzzy logic controller receiving n E input variables and outputting n LW output variables, comprising: fuzzification circuit means having a memory for storing membership functions of the input variables, said fuzzification circuit means generating u n E combinations of first numbers for linguistic values of input variables addressed by externally supplied distinct values of the n E input variables and values of membership functions of the linguistic values of the input variables addressed by the externally supplied distinct values of the n E input variables, whereby a maximum of u membership functions respectively overlap and, as a result thereof, a maximum of u linguistic values are simultaneously addressed by a distinct value of the input variables; rule decoder means having a linquistic protocol and for forming, in conformity with the linguistic protocol stored therein, second numbers of linguistic values of the output variables from the u n E combinations of the first numbers, the linguistic protocol being IF/THEN rules and the u n E combinations of the first numbers being an IF part of said rules and the second members being a THEN part of said rules, said rule decoder means having means for forming selection signals for definition of the input variables affected by a respective rule; rule evaluation circuit means for generating a weighting signal for each of n LW linguistic values of the output variables from the second numbers of linguistic values of the output variables and from the values of the membership functions of respective linguistic values of the output variables by means of at least one circuit for the formation of a nondistinct operation, said selection signals being supplied to the rule evaluation circuit means; inference circuit means having a memory and at least one further circuit for the formation of a nondistinct operation, and said inference circuit means forming signals for nondistinct unification sets at respective locations from the weighting signals for all locations of an area of distinct values of the output variables; a defuzzification circuit means having at least one evaluation circuit, said defuzzification circuit means generating distinct, externally available values of the output variables from the signals for the nondistinct unification sets of all locations of the area of distinct values of the output variables; and sequencer means for effecting that, after the u n E combinations of first numbers, the weighting signals for the n LW linguistic values of the output variables are calculated once for every set of distinct values of the n E input variables, said weighting signals being used in the inference circuit for all locations of the area of distinct values of the output variables, the sequencer means also effecting that nondistinct unification sets for already calculated locations of the area of distinct values of the output variables are simultaneously further-processed while a respective signal for the nondistinct unification set for a location of the area of distinct values of the output variables is calculated in the inference circuit means.

Claim 13

The fuzzy logic controller according to claim 10, wherein the rule evaluation circuit means has circuit means for formation of a limited sum and maximum circuit means that serve as first and second circuits, respectively, for the formation of a nondistinct operation; and wherein, alternatively, one of the first and second circuits is used for the formation of the weighting signals dependent on an externally supplied rule evaluation select signal.

Claim 14

The fuzzy logic controller according to claim 10, wherein the inference circuit means has circuit means for the formation of a limited sum and a maximum circuit means that serve as first and second, further circuits for the formation of a nondistinct operation; and wherein, alternatively, one of the first and second circuits is used for the formation of the signals for the nondistinct unification set at respective locations dependent on an externally suppliable inference select signal.

Claim 15

The fuzzy logic controller according to claim 10, wherein the defuzzification circuit means has a circuit for calculation of a center of gravity and a circuit for calculation of a mean of maxima that serve as first and second evaluation circuits; and wherein, alternatively, one of the first and second evaluation circuits is used for the formation of the distinct values of the output variables dependent on an externally suppliable defuzzification select signal.

Claim 16

The fuzzy logic controller according to claim 10, wherein the rule decoder means is programmable logic.

Claim 17

The fuzzy logic controller according to claim 10, wherein at least one of the memory for the input variables and the memory for the output variables is a random-access memory.

Claim 18

The fuzzy logic controller according to claim 10, wherein the fuzzy logic controller together with its memory for the input variables and its memory for the output variables is a component part of a conventional microcontroller.

Claim 19

A fuzzy logic controller, said fuzzy logic controller receiving n E input variables and outputting n LW output variables, comprising: fuzzification circuit means having a memory for storing membership functions of the input variables, said fuzzification circuit means generating u n E combinations of first numbers for linguistic values of input variables addressed by externally supplied distinct values of the n E input variables and values of membership functions of the linguistic values of the input variables addressed by the externally supplied distinct values of the n E input variables, whereby a maximum of u membership functions respectively overlap and, as a result thereof, a maximum of u linguistic values are simultaneously addressed by a distinct value of the input variables; rule decoder means having a linguistic protocol and for forming, in conformity with the linguistic protocol stored therein, second numbers of linguistic values of the output variables from the u n E combinations of the first numbers, the linguistic protocol being IF/THEN rules and the u n E combinations of the first numbers being an IF part of said rules and the second members being a THEN part of said rules; rule evaluation circuit means for generating a weighting signal for each of n LW linguistic values of the output variables from the second numbers of linguistic values of the output variables and from the values of the membership functions of respective linguistic values of the output variables by means of circuit means for formation of a limited sum and maximum circuit means that serve as first and second circuits, respectively, for the formation of a nondistinct operation, one of the first and second circuits being used for the formation of the weighting signals dependent on an externally supplied rule evaluation select signal; inference circuit means having a memory and the inference circuit means having circuit means for the formation of a limited sum and a maximum circuit means that serve as first and second, further circuits for the formation of a nondistinct operation, one of the first and second circuits being used for formation of signals for a nondistinct unification set at respective locations dependent on an externally suppliable inference select signal, and said inference circuit means forming signals for nondistinct unification sets at respective locations from the weighting signals for all locations of an area of distinct values of the output variables; a defuzzification circuit means having a circuit for calculation of a center of gravity and a circuit for calculation of a mean of maxima that serve as first and second evaluation circuits, one of the first and second evaluation circuits being used for the formation of distinct values of the output variables dependent on an externally suppliable defuzzification select signal, said defuzzification circuit means generating the distinct, externally available values of the output variables from the signals for the nondistinct unification sets of all locations of the area of distinct values of the output variables; and sequencer means for effecting that, after the u n E combinations of first numbers, the weighting signals for the n LW linguistic values of the output variables are calculated once for every set of distinct values of the n E input variables, said weighting signals being used in the inference circuit for all locations of the area of distinct values of the output variables.

Claim 20

A fuzzy logic controller, said fuzzy logic controller receiving n E input variables and outputting n LW output variables, comprising: fuzzification circuit means having a memory for storing membership functions of the input variables, said fuzzification circuit means generating u n E combinations of first numbers for linguistic values of input variables addressed by externally supplied distinct values of the n E input variables and values of membership functions of the linguistic values of the input variables addressed by the externally supplied distinct values of the n E input variables, whereby a maximum of u membership functions respectively overlap and, as a result thereof, a maximum of u linguistic values are simultaneously addressed by a distinct value of the input variables; rule decoder means having a linguistic protocol and for forming, in conformity with the linguistic protocol stored therein, second numbers of linguistic values of the output variables from the u n E combinations of the first numbers, the linguistic protocol being IF/THEN rules and the u n E combinations of the first numbers being an IF part of said rules and the second members being a THEN part of said rules; rule evaluation circuit means for generating a weighting signal for each of n LW linguistic values of the output variables from the second numbers of linguistic values of the output variables and from the values of the membership functions of respective linguistic values of the output variables by means of circuit means for formation of a limited sum and maximum circuit means that serve as first and second circuits, respectively, for the formation of a nondistinct operation, one of the first and second circuits being used for the formation of the weighting signals dependent on an externally supplied rule evaluation select signal; inference circuit means having a memory and the inference circuit means having circuit means for the formation of a limited sum and a maximum circuit means that serve as first and second, further circuits for the formation of a nondistinct operation, one of the first and second circuits being used for formation of signals for a nondistinct unification set at respective locations dependent on an externally suppliable inference select signal, and said inference circuit means forming signals for nondistinct unification sets at respective locations from the weighting signals for all locations of an area of distinct values of the output variables; a defuzzification circuit means having a circuit for calculation of a center of gravity and a circuit for calculation of a mean of maxima that serve as first and second evaluation circuits, one of the first and second evaluation circuits being used for the formation of distinct values of the output variables dependent on an externally suppliable defuzzification select signal, said defuzzification circuit means generating the distinct, externally available values of the output variables from the signals for the nondistinct unification sets of all locations of the area of distinct values of the output variables; and sequencer means for effecting that, after the u n E combinations of first numbers, the weighting signals for the n LW linguistic values of the output variables are calculated once for every set of distinct values of the n E input variables, said weighting signals being used in the inference circuit for all locations of the area of distinct values of the output variables.

Claim 21

The fuzzy logic controller according to claim 17, wherein the sequencer means also effects that nondistinct unification sets for already calculated locations of the area of distinct values of the output variables are simultaneously further-processed while a respective signal for the nondistinct unification set for a location of the area of distinct values of the output variables is calculated in the inference circuit means.

Claim 22

The fuzzy logic controller according to claim 17, wherein selection signals for definition of the input variables affected by a respective rule are additionally formed in the rule decoder means, wherein said selection signals are supplied to the rule evaluation circuit means, and wherein the rule decoder means is programmable logic.

Claim 23

The fuzzy logic controller according to claim 17, wherein at least one of the memory for the input variables and the memory for the output variables is a random-access memory.