The Modeling Method Research Based On T-S Model

Building object model is the ground basis of designing a control system. But modeling nonlinear system is the obstacle to model the whole complicated system. The current researches have proved that T-S fuzzy model is the effective approach to build the model for this kind of system. Compared with all kinds of modeling methods based on T-S fuzzy model, we suggest the well-developed ANFIS modeling method and fuzzy clustering modeling method. In the paper, we give the detailed introduction of T-S fuzzy model, ANFIS modeling method and fuzzy clustering modeling method.This paper focuses on the mixed learning algorithm that is used by T-S fuzzy model learning. Besides this, we also attach importance to the ANFIS modeling method, the different model structure of fuzzy clustering modeling method and the variant approach of these two methods to segment the input space. Aiming at the problem caused by the traditional modeling method based on T-S fuzzy model that it does not take the correlativity of training data into account, we present an effective approach to remove the correlativity of training data by using optimized orthogonal transformation. Then we can provide the method to improve the traditional way. In the end, we guarantee the effectiveness of new method through simulation examples.