Fuzzy Model Structure Analysis and Simplification Based on Singular Value Decomposition

Using matrix singular value decomposition, the structure of a class of fuzzy models and the relations between singular values and fuzzy rules are studied. A rate of the fuzzy rule accumulative contribution is proposed to guide the process of the fuzzy model simplification in order to reach a balance between model fitting accuracy and model structure simplicity. The theoretical analysis and simulation results show that it is feasible and practicable.