GANFIS Fault Tolerant Control Using Linear Transformations of the Evolutionary Fitness Function

Fault detection, isolation and correction of nonlinear systems continue to be important problems to be addressed due to the increased complexities of more advanced systems. This paper addresses the issue of fault tolerant, robust control design; a generalized adaptive fuzzy neural network-based inference system is used for the compensation due to faults and evolutionary techniques are used to tune the fuzzy neural inference parameters. A linear transformation of the standard squared-error fitness is used here. This strategy can be used in a closed-loop feedback control structure to improve performance. This is illustrated on a discrete-time model of a nonlinear system and shows that proposed method provides fault compensation for nonlinear systems.