Universal approximation by hierarchical fuzzy system with constraints on the fuzzy rule (Q1855480)

This paper presents a special hierarchical fuzzy system where the outputs of the previous layer are not used in the IF-parts, but used only in the THEN-parts of the fuzzy rules of the current layer. The proposed scheme can be shown to be a universal approximator to any continuous function on a compact set if complete fuzzy sets are used in the IF-parts of the fuzzy rules with singleton fuzzifier and center average defuzzifier. The conventional HFS (Hierarchical Fuzzy System) does reduce the number of fuzzy rules dramatically, but it is difficult to design because intermediate output variables with less or no physical meaning are used as input variables for the fuzzy logic units at the next layer. In contrast, the proposed HFS uses only the original input variables with clear physical meaning in all of the IF-parts of the fuzzy rules, thereby rendering the involved fuzzy rules easy to interpret and design. Any real continuous function on a compact set is proved to be approximated by the proposed HFS to any degree of accuracy. From the simulation of a ball and beam control system, it is demonstrated that the proposed scheme approximates with good accuracy the model nonlinear controller with fewer fuzzy rules than the centralized fuzzy system, and its control performance is comparable to that of the nonlinear controller.