Hybrid-fuzzy controller optimization via semi-parallel GA for servomotor control

Servomotor uses feedback controller to control either the speed or the position or both. This paper discusses the performance comparisons of a modified genetic algorithm, named as the semi-parallel operation genetic algorithm (SPOGA) and the conventional genetic algorithm (GA), in optimizing the I/O scale factors, membership functions, and rules of a hybrid-fuzzy controller. Singleton fuzzification is used as a fuzzifier with seven membership functions for both input and output of the controller, whilst center of average is used as a defuzzifier. A 21-bit-30-population is used in SPOGA for both I/O scales and for membership functions. Two control modes are applied in cascaded: position and speed. Both the simulation and practical experiment results show that fuzzy-logic parallel integral controller (FLIC) with SPOGA-optimized is better as compared to FLIC with GA-optimized and also the non-optimized FLIC, FLC, and PI in terms of performance and the reduction of the number of test runs for the optimization.