Abstract:

In order to improve the accuracy of the fan fault diagnosis, the author presented a new method of fan fault diagnosis which was the combination of evidence theory and support vector machine. Firstly, WignerVille spectrum entropy was extracted from the vibration signal as characteristic of fan fault diagnosis. Secondly, subclassifier of fan fault diagnosis was established by using different kernel function support vector machines. Finally, the output results of subclassifier were fused by DS evidence theory, and the performance was simulated and tested. The experimental results show that the proposed method can make full use of all fault information, and the diagnostic results are closer to the expected value, and the diagnosis effect is better than that of other fan fault diagnosis methods.

Key words: fan fault, feature extraction, evidence theory, support vector machine