Data-Driven Arithmetic Fuzzy Control Using the Distending Function

Abstract

Data-driven fuzzy control techniques suffer from the flat structure problem, i.e. the number of fuzzy rules grows exponentially as the input dimension increases. The consequence is greater complexity and poorer interpretability. In this article, we present a solution to the above-mentioned problem by proposing a novel data-driven fuzzy controller. Its unique features are: (1) A new type of membership function is used; (2) It helps to identify few and important fuzzy rules; (3) These rules cover the whole input space; (4) Dombi operators (usually the conjunctive) are employed to generate a higher dimensional control surface; (5) The designed fuzzy controller is based on fuzzy arithmetic operations; (6) Defuzzification is single step calculation. Due to the small number of fuzzy rules, the complexity of the fuzzy model decreases and it becomes interpretable. The effectiveness of the proposed scheme is demonstrated using the data-driven based control of vehicle lateral dynamics.