The limited capability of conventional controllers to control complicated, dynamical, systems has motivated research into so-called intelligent control systems. In this project, a genetic algorithm is used to tune a fuzzy logic controller by finding a set of controller coefficients that minimize the tracking error and control effort. The principle assumption in this application is that traditional analysis has not been performed in the design of the controller. The control system will learn how to control the nonlinear system by learning the system''s characteristics in real time. The genetic algorithm has been applied in the optimization of various aspects of these intelligent controllers. Work in the area of GAs applied to fuzzy control is broadly split into two categories: tuning of the membership functions and elicitation of the rulebase in by varying the rulebase parameters. Most of this work has been performed offline. In our system, the tuning of the fuzzy logic controller takes place on line, while the system is operating.

Michael R. McKee, MSEE: Studied control systems at California State University, Northrigde, BS Physics: Studied general physics and plasma physics at University of California at Los Angeles. Currently, Senior Engineer in the Measurement Systems Group / JPL.