Fuzzy Space Partitioning Based on Hyperplanes Defined by Eigenvectors for Takagi-Sugeno Fuzzy Model Identification

Abstract

This article presents a novel method for fuzzy space partitioning and the identification of Takagi-Sugeno fuzzy models. The novelty is in its region-splitting mechanism and membership function definition, which is based on hyperplanes. The proposed algorithm introduces a concept of principal component analysis to define the hyperplanes that split the problem space and uses the distances to these hyperplanes as metrics instead of center-oriented clusters. In contrast with many other methods, the presented method delivers reproducible results and has an easy tuning procedure. The performance is illustrated with analytical examples, benchmark problems from the literature, and real-process data. The obtained results are very promising; however, as with most learning methods, the results depend on the data distribution and input variable selection.