On robustness of evolutionary fuzzy control systems

The effect of evolutionary optimization of fuzzy control systems on their robustness has been studied through simulation. It is shown here that one of the side effects of the evolutionary fuzzy control systems may be decreasing robustness in price of increasing optimality and performance. In this paper the above problem has been studied for a simple case study. A fuzzy control system is first designed for a back truck parking problem and robustness of this controller has been tested under changes in system parameters. The fuzzy controller is then optimized using a simple Genetic Algorithm and it is shown that the evolutionary fuzzy controller has better performance than the initial non-model based fuzzy controller. It is shown that the performance of the optimized fuzzy controller is more dependent on the nominal model of the system and hence has lower robustness.