Autotuning for fuzzy-PI control using genetic algorithm

The paper presents a methodology in applying genetic search techniques to automatically select a proper triangular membership function for fuzzy logic based PI controllers. The approach produces a novel type of fuzzy-PI controllers with guaranteed stability property and the desired performance specified by the more general performance criterion. The control approach is suitable for linear systems as well as a class of nonlinear unstable plant studied by Tseng et al. Numerical simulation results show that the better membership functions can be obtained and the resultant hierarchical fuzzy-PI controller outperforms the fix-valued PI controller.