Establishing interpretable fuzzy models from numeric data

Abstract

In this paper, a rule-base self-extraction and simplification method is proposed to establish interpretable fuzzy models from numerical data. A fuzzy clustering technique is used to extract the initial fuzzy rule-base. The number of fuzzy rules is determined by the proposed fuzzy partition validity index. To reduce the complexity of fuzzy models without decreasing the model accuracy significantly, some approximate similarity measures are presented and a parameter fine-tuning mechanism is introduced to improve the accuracy of the simplified model. Using the proposed similarity measures, the redundant fuzzy rules are removed and similar fuzzy sets are merged to create a common fuzzy set. The simplified rule base is computationally efficient and linguistically tractable. The approach has been successfully applied to non-linear function approximation and mechanical property prediction for hot-rolled steels.