Design of neuromorphic fuzzy controllers
Duc Truong Pham, Derviş Karaboğa · 2002
The paper introduces a general neural model for a SISO fuzzy logic controller (FLC). An FLC represented in the form of a neural network can be trained using a genetic algorithm (GA). This enables the simultaneous determination of the membership functions for the fuzzy input variable, the quantisation levels for the output variable and the elements of the relation matrix of the FLC. The paper presents simulation results for the control of a time delayed second order system which show the fast and accurate performance of a GA trained neuromorphic FLC.>