Rule Extraction through Self-Organizing Map for a Self-Tuning Fuzzy Logic Controller

Complex, imprecise, ill-defined and uncertain systems can be identified by fuzzy modeling. The identification of such a linguistic model consists of two parts; structure identification and parameter estimation. In this paper, we proposed Self-Organizing Map (SOM) clustering technique for structure identification. Initial parameters of the input-output membership functions (MFs) will be estimated from the clustering results. In the process of proto-type generation, the gradient descent method is used to refine and tune the parameters effectively. The proposed scheme has been successfully tested to identify the rules required to realize the gain factor of self-tuning fuzzy PI controller, which is highly non-linear in nature. Here, the system identification method is integrated with rule extraction method in such a way that it can pick up required rules from any processes. Finally, examples of various systems are given to illustrate the effectiveness of the proposed approach.