Fuzzy Controller Design Based On Clustering Algorithm

In the process of modern industrial control, fuzzy control is widely used in complex process and object control for its virtues such as easy realization, robust. Fuzzy rule base, as crucial components of the control systems, are good or not will directly affect the control performance. The conventional design method is by summarizing the expert experiences to produces the fuzzy control rule, such fuzzy controller relies on expert's subjective experience intensely, and the fuzzy rules cannot reflect the substantive characteristics of the control system completely. So how to get the fuzzy rules automatically are always important for the fuzzy control theory research.Firstly, this paper describe the data preprocessing method, analysis the vacant data pack algorithm, and carry out the data pack via spline insertion. Also initialize the FCM prototype, solved the problem of sensitively to the initial value of this algorithm. Then, use the nearest clustering algorithm and the FCM clustering algorithm separately to generate the fuzzy rules automatically, make one rule for every group of data, this method can generate fuzzy rules specially, and reduces the redundant rules. We combine FCM clustering algorithm with Genetic Algorithm, figure out the FCM by GA, it is a good mentality for the problem that is sensitive to the initial value. Finally optimize the fuzzy controller by GA, through making reasonable logical coding rule, choose the appropriate genetic operator, enhanced the rules quality.The paper research fuzzy rules automatically generated by clustering algorithm, solved the problem of hard to determine the fuzzy rules and the subjective. This method can make different input variables have different membership degree, and calculate the FCM with Genetic Algorithm, enhanced the algorithm efficiency and the rule quality. The experiment results illustrate the feasibility and the effectiveness.