Rough Set Data Analysis System and Its Applications in Machinery Fault Diagnosis
The characteristics of fault diagnosis are as follows. First, features extraction is the key of improving diagnosis efficiency and correct rate. Secondly, fault diagnosis method based on rule reasoning has a wide application, but rule acquisition is one of the bottlenecks. Thirdly, rule modification is a key question of solving the real-time rule acquisition in the dynamic environments, and a primary question of knowledge base modification of expert system, etc. In this paper, Rough Set Theory (RST) was used to solve the key problems of machinery fault diagnosis, and a Rough Set Data Analysis System (RSDAS) was developed. RSDAS was used to implement rule generation automation & rule modification based on RST such as indiscernibility relation and knowledge reduction method, depicted importance of different attributes in knowledge representation, and reduced knowledge representation space. The method of fault diagnosis using RSDAS was summarized. The experiment results approved the feasibility and the high precision of RSDAS. Therefore, we can use RADAS to machinery fault diagnosis.
Xing Ai, Jianfeng Li and Chuanzhen Huang
L.N. Hao et al., "Rough Set Data Analysis System and Its Applications in Machinery Fault Diagnosis", Materials Science Forum, Vols. 471-472, pp. 850-854, 2004