T-S fuzzy system identification based on support vector machine

Abstract

There are some problems in fuzzy system for modeling and identification, such as complexity of model construction, curse of dimensionality, poverty of generalization and error of real-time. To deal with these problems, support vector mechanism (SVM) for fuzzy system modeling has been introduced in this paper. And then the parameters have been optimized by error back-propagation training algorithm (BP algorithm). Experimental results demonstrate the effectiveness of the method.