Genetic Algorithm Tuned Fuzzy Logic Controller for Rotary Inverted Pendulum

Abstract

In this study, a Genetic Algorithm (GA) is proposed to search for the optimal input membership functions of the fuzzy logic controller. With the optimal membership function, the fuzzy logic controller can efficiently control a rotary inverted pendulum. The advantage of the proposed method is tuning the parameters of membership functions automatically rather than tuning them manually. In genetic algorithm, these parameters are converted to a chromosome which is encoded into a binary string. Because the membership functions are symmetric to zero, the length of each chromosome could be reduced by half. The computation time will also be shorter with the shorter chromosomes. Moreover, the roulette wheel selection is chosen as reproduction operator and one-point crossover operator and random mutation operator are also used. After the genetic algorithm completes searching for optimal parameters, the optimal membership function will be introduced to the fuzzy logic controller. Finally, simulation results show that the proposed GA-tuned fuzzy logic controller is effective for the rotary inverted pendulum control system with robust stabilization capability.