On improving fuzzy variable-structure controller using adaptive neural networks

To overcome the deficiencies of the traditional fuzzy variable-structure controller in lack of learning capability and high frequency flutter under steady-state condition, the improved fuzzy variable-structure controller based on adaptive neural networks was proposed. The learning capability of adaptive neural element was used for self-learning of the fuzzy variable-structure control rules, and the flutter phenomenon was removed by composition of fuzzy reasoning. The simulation results have shown that the improved controller has fast dynamic response, small overshooting and strong robustness.