Elastic fuzzy logic system

Claim 3

The apparatus according to claim 1 wherein functions f and g are chosen such that said rule memory stores and said processing unit processes said plurality of if-then rules according to R.sub.i =γ.sub.i0 μ.sub.i1.sup.γ.sbsp.i,1 *μ.sub.i2.sup.γ.sbsp.i,2 * . . . *μ.sub.i,m.sup.γ.sbsp.i,m where at least one of (γ i0 through γ i ,m) is not 1.0.

Claim 4

The apparatus according to claim 1, further comprising a defuzzification device for defuzzifying the output of the processing unit.

Claim 5

The apparatus according to claim 2, further comprising a defuzzification device for defuzzifying the output of the processing unit.

Claim 6

The apparatus according to claim 2, wherein said data entered by said input device further comprises corresponding output values for said at least one input device, and wherein the rule memory comprises: means for initially setting γ i .sbsb.0 through γ i ,m to corresponding initial values; means for adapting γ i .sbsb.o through γ i ,m using a learning process based on said at least one input value and said corresponding output values.

Claim 7

The apparatus according to claim 5, wherein the means for initially setting comprises a means for setting γ i .sbsb.0 through γ i ,m to 1.0.

Claim 8

The apparatus according to claim 5, wherein the means for initially setting comprises a means for setting γ i .sbsb.0 through γ i ,m randomly.

Claim 9

The apparatus according to claim 5, wherein the means for adapting comprises means for updating γ i .sbsb.0 through γ i ,m using a neural network learning process.

Claim 10

The apparatus according to claim 5, wherein the means for adapting comprises means for updating γ i .sbsb.0 through γ i ,m using back propagation.

Claim 11

The apparatus according to claim 5, wherein the means for adapting comprises means for updating γ i .sbsb.0 through γ i ,m using a dual subroutine.

Claim 14

A computer program product as in claim 11, wherein the second computer code device is configured to select functions f and g such that said plurality of rules are stored in a form R.sub.i =γ.sub.i0 *μ.sub.i2.sup.γ.sbsp.i,1 *μ.sub.i1.sup.γ.sbsp.i,2 * . . . *μ.sub.i,m.sup.γ.sbsp.i,m where at least one of (γ i0 through γ i ,m) is not 1.0.

Claim 15

The method according to claim 11, further comprising the step of defuzzifying the output of the outputting step.

Claim 16

A method as in claim 11, the step of storing said plurality of if-then rules comprises choosing functions f and g such that said plurality of rules are stored in a form R.sub.i =γ.sub.i0 *μ.sub.i1.sup.γ.sbsp.i,1 *μ.sub.i2.sup.γ.sbsp.i,2 * . . . *μ.sub.i,m.sup.γ.sbsp.i,m where at least one of (γ i0 through γ i ,m) is not 1.0.

Claim 17

The method according to claim 14, further comprising the step of defuzzifying the output of the outputting step.

Claim 18

The method according to claim 11, further comprising the steps of: inputting corresponding output values for the input values received in the step of inputting data; setting γ i .sbsb.0 through γ i ,m to initial values initially; adapting γ i .sbsb.0 through γ i ,m using a learning process based on said input values and said corresponding output values.

Claim 19

The method of claim 16, wherein the step of setting γ i .sbsb.0 through γ i ,m initially comprises setting γ i .sbsb.0 through γ i ,m to 1.0.

Claim 20

The method of claim 16, wherein the step of setting γ i .sbsb.0 through γ i ,m initially comprises setting γ i .sbsb.0 through γ i ,m randomly.

Claim 21

The method according to claim 16, wherein the step of adapting comprises updating γ i .sbsb.0 through γ i ,m using a neural network learning process.

Claim 22

The method according to claim 16, wherein the step of adapting comprises updating γ i .sbsb.0 through γ i ,m using back propagation.

Claim 23

The method according to claim 16, wherein the step of adapting comprises updating γ i .sbsb.0 through γ i ,m using a dual subroutine.

Claim 24

The method according to claim 16, further comprising the steps of: reporting to a user the adapted values of γ i .sbsb.0 through γ i ,m.sbsb.0, and updating at least one of the plurality of membership functions based on the updated values of γ i .sbsb.0 through γ i ,m.

Claim 25

In an apparatus for fuzzy control, including a membership memory for storing a plurality of membership functions for fuzzy control using at least one input variable and output variable; a rule memory for storing a plurality of if-then rules; an input device for entering data comprising at least one input value; a processing unit for receiving said at least one input value from said input device, for retrieving at least one membership function of said plurality of membership functions from said membership memory, for retrieving at least one rule of said plurality of if-then rules from said rule memory and for producing an output representing a degree to which the at least one rule applies to said at least one input value using said at least one membership function, said at least one input value, and said at least one rule, the improvement comprising: storing said if-then rules in a form R=f(γ.sub.0,.sub.i.sup.π g(γ.sub.i,μ.sub.i)), where each μ i is one of the plurality of membership functions, where f and g are differentiable, and for which there exists a γ 0 which causes R to equal zero, and for which there exists a base value for γ i which causes clause i to effectively be removed from the rule.

Claim 26

In an apparatus for fuzzy control, including a membership memory for storing a plurality of membership functions for fuzzy control using at least one input variable and output variable; a rule memory for storing a plurality of if-then rules; an input device for entering data comprising at least one input value; a processing unit for receiving said at least one input value from said input device, for retrieving at least one membership function of said plurality of membership functions from said membership memory, for retrieving at least one rule of said plurality of if-then rules from said rule memory and for producing an output representing a degree to which the at least one rule applies to said at least one input value using said at least one membership function, said at least one input value, and said at least one rule, the improvement comprising: storing said if-then rules in a form R=f(γ.sub.0,.sub.i.sup.π g(γ.sub.i,μ.sub.i)), where each μ i is one of the plurality of membership functions, where f and g are differentiable, and for which there exists a γ 0 which causes R to equal zero, and for which there exists a base value for γ i which causes clause i to effectively be removed from the rule.

Claim 27

The improvement of claim 23, wherein said if-then rules are stored in a form R.sub.i =γ.sub.io *μ.sub.i1.sup.γ.sbsp.i,1 *μ.sub.i2.sup.γ.sbsp.i,2 * . . . *μ.sub.i,m.sup.γ.sbsp.i,m.

Claim 28

A computer program product comprising: a computer storage medium and a computer program code mechanism embedded in the computer storage medium for causing a computer to implement a fuzzy controller having a membership memory, a multiplier factor memory and a rule memory, the computer program code mechanism comprising: a first computer code device configured to store in said membership memory a plurality of membership functions for fuzzy control based on at least one input variable and an output variable; a second computer code device configured to store in said rule memory a plurality of if-then rules in a form R=f(γ.sub.0,.sub.i.sup.π g(γ.sub.i,μ.sub.i)), where each μ i is one of the plurality of membership functions, where f and g are differentiable, and for which there exists a γ 0 which causes R to equal zero, and for which there exists a base value for γ i which causes clause i to effectively be removed from the rule; a third computer code device configured to input data for said at least one input variable, said data comprising input values; a fourth computer code device configured to select at least one membership function from the plurality of membership functions; a fifth computer code device configured to select at least one if-then rule from the plurality of if-then rules; and a sixth computer code device configured to output an output corresponding to a degree to which the at least one if-then rule applies to the at least one input variable by using the at least one membership function, and the at least one if-then rule.

Claim 29

A computer program product comprising: a computer storage medium and a computer program code mechanism embedded in the computer storage medium for causing a computer to implement a fuzzy controller having a membership memory, a multiplier factor memory and a rule memory, the computer program code mechanism comprising: a first computer code device configured to store in said membership memory a plurality of membership functions for fuzzy control based on at least one input variable and an output variable; a second computer code device configured to store in said rule memory a plurality of if-then rules in a form R=f(γ.sub.0,.sub.i.sup.π g(γ.sub.i,μ.sub.i)), where each μ i is one of the plurality of membership functions, where f and g are differentiable, and for which there exists a γ 0 which causes R to equal zero, and for which there exists a base value for γ i which causes clause i to effectively be removed from the rule; a third computer code device configured to input data for said at least one input variable, said data comprising input values; a fourth computer code device configured to select at least one membership function from the plurality of membership functions; a fifth computer code device configured to select at least one if-then rule from the plurality of if-then rules; and a sixth computer code device configured to output an output corresponding to a degree to which the at least one if-then rule applies to the at least one input variable by using the at least one membership function, and the at least one if-then rule.