Constructing Interpretable Fuzzy Model Based on Reduction Methodology

Authors:	Zongyi Xing, Nanjing University of Science and Technology, China Weilia Hu, Nanjing University of Science and Technology, China Liminb Jia, Beijing Jiaotong University, China
Topic:	3.2 Cognition and Control ( AI, Fuzzy, Neuro, Evolut.Comp.)
Session:	Intelligent Modelling and Identification II
Keywords:	fuzzy modelling; fuzzy sets; input estimation; fuzzy model; genetic algorithms

Abstract

A systematic approach for constructing interpretable fuzzy model based on reduction methodology is proposed. Fuzzy clustering algorithm, combined with least square method, is used to identify initial fuzzy model with overestimated rule number. Orthogonal least square algorithm and similar fuzzy sets merging are then applied to remove redundancy of the fuzzy model. In order to obtain high accuracy, yet preserving interpretability, a constrained real coded genetic algorithm is utilized to optimize reduced fuzzy model. The proposed method was applied to automobile MPG prediction, and results show its validity.