Fuzzy model identification and fuzzy controller design based on data characteristic

Wenbiao Zhu,Zengqi Sun
2009-01-01
Abstract:A fuzzy model identification method based on data characteristic is presented for complex processes which are difficult to be mathematically modeled. First, the domain of discourse of input variables is divided according to the changing degree of the process output while the input variables change. Then, the total number and the premise parameters of the fuzzy rules are determined based on the division. Finally, the fuzzy model can be expressed as a feedfoward fuzzy neural network, so the BP algorithm is applied to obtain the consequent parameters of the fuzzy rules. The fuzzy controller of a flexible link robot is designed based on the presented fuzzy model identification method, and the corresponding fuzzy control experiment is performed. The experimental results show that the satisfactory performance is achieved by using the presented fuzzy model identification and fuzzy control method.
What problem does this paper attempt to address?