A Diagnostic Method of Rotating Machinery Based on Rough Sets Theory and Pattern Recognition
Wu Xing · Mechanical Science and Technology · 2004
A diagnostic method of rotating machinery based on rough sets theory and pattern recognition is proposed. This method includes two processes: one is pattern learning, and the other is pattern recognition. The standard diagnostic rules, i.e. standard patterns, are obtained by rough-set-technique-based learning method from standard fault samples and the diagnostic conclusions are produced by matching diagnostic objects with standard diagnostic rules. The proposed pattern learning method takes into account the reduplicated and conflicting objects in decision table and makes the obtained rules cover all the learning objects. In the pattern recognition, the matching degree of condition attributes of new objects with standard diagnostic rules, the belief degree of those rules and the conclusion threshold are under consideration for conclusion and its belief degree. Consequently, the conclusion based on the proposed method is objective. This method has been used in the diagnosis of rotating machinery and the result is satisfactory.