Automatically Constructed Fuzzy Rule-Based Pattern Classification Systems for Fault Diagnosis

자동 구축 퍼지 규칙기반 패턴 인식 시스템에 의한 고장진단 시스템의 구현

This paper presents the automatic construction of fuzzy rule-based systems for diagnosing the faults of complex systems. Generally, fuzzy systems work well when we can use expert's experience to articulate fuzzy IF-THEN rules and memberships for fuzzy sets. When we cannot do this, we should generate the fuzzy rules and membership functions for fuzzy sets directly from experimental data. In this paper, we propose a new method on how to extract fuzzy sets and fuzzy rules. We also introduce an efficient fine-tunning algorithm of the parameters of membership functions.