Condition Diagnosis for Rotating Machinery Using Support Vector Machines and Symptom Parameters in Frequency Domain

Up to now, many condition diagnosis methods based on the traditional artificial intelligence, such as neural networks (NN), genetic algorithms (GA), etc., have been proposed in the field of condition diagnosis for rotating machinery. These methods depend on the assumption that the number of samples tends to infinity, and also require a large amount of training samples and highly sensitive symptom parameters (SPs). However, as the satisfied samples cannot be easily acquired from a real plant and SPs are not so highly sensitive as supposed to be. In many cases of condition diagnosis for rotating machinery, the intelligent methods, such as neural networks, genetic algorithms, etc., often cannot converge when learning. In order to solve these problems, a new condition diagnosis method using support vector machines (SVWs) is proposed in this paper. The practical examples of diagnosis for rotating machinery are shown to verify the efficiency of the proposed method.