Optimization of Fuzzy Membership Function based on the NCOS function method
Jingying Li, Xingjian Jing, Zhengchao Li, Xianlin Huang · IECON 2020 The 46th Annual Conference of the IEEE Industrial Electronics Society · 2020
This paper investigates the parameter optimization problem of membership functions for fuzzy-model-based control under imperfect premise matching. A novel frequency domain algorithm which can clearly relate the membership function parameters to the targeted control performance is developed, and consequently an optimization approach to the membership function parameters is then established based on the resulting nonlinear characteristic output spectrum function (nCOS). Compared to traditional search-based optimization approach, this method can give a more detailed result with less time consuming and an in-depth understanding of nonlinear influence rather than just optimal results. With this novel method, performance of the fuzzy-model-based controller is further enhanced. Finally, the fuzzy membership functions optimization method is applied to nonlinear systems to obtain improved system performance.