A dynamic T-S fuzzy systems identification algorithm based on sparsity regularization

Fuzzy systems identification suffers from "rules explosion", i.e., the number of fuzzy rules grows exponentially with the increase of the dimension of the input variable. In this paper, a dynamic algorithm is exploited to address T-S fuzzy systems identification on the basis of sparsity regularization. With a dynamic increase of fuzzy rules, this method automatically extracts fuzzy rules' antecedent part in a way of iterative vector quantization clustering and estimates the parameters of fuzzy rules' consequent part on the basis of sparsity regularization. In such a way, a minimal number of fuzzy rules and nonzero consequent parameters can be guaranteed in T-S fuzzy systems identification. Finally, some numerical experiments on a well-known benchmark dataset are carried out to verify the effectiveness of the proposed approach.