Abstract
Clustering techniques for the generation of fuzzy models have been used and have shown promising results in many applications involving complex data. This chapter proposes a new incremental clustering technique to improve the discovery of local structures in the obtained fuzz ' v models. This clustering method is evaluated oil two data sets and the results are compared with,h the results of other clustering methods. The proposed clustering approach is applied for nonlinear Takagi-Sugeno (TS) fuzzy modeling. This incremental clustering procedure that generates clusters that, are used to form the fuzzy rule antecedent part in online mode is used as a first stage of the learning process.