Fuzzy supervisory PI controller using hierarchical genetic algorithms

This paper presents a fuzzy supervisory PI controller using a hierarchical genetic algorithm (HGA) which is an optimization algorithm used in defining the optimal number and shape of membership functions, and fuzzy rules. The fuzzy supervisory PI controller is designed and simulated on a heat exchanger system. Sum of square error and sum of absolute error are used as objective functions. Simulation results are compared with the PI controller using Astrom and Hagglund tuning. Improved performances can be noticed.