Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
406838 | Neurocomputing | 2013 | 10 Pages |
This paper presents a new, intelligent diagnostic method for identifying structural faults in rotating machinery based on distinctive frequency components (DFCs) and support vector machines (SVMs) under varied operating conditions. This method consists of three stages. First, when investigating and comparing the spectrum feature of structural faults in the most salient frequency band, the DFCs can be detected and extracted. Second, when analyzing the common DFCs from various operating conditions, the DFCs are normalized on the universal standard to reduce the difference. Then, the optimal DFC area of any state under various operating conditions can be detected using probability theory. Finally, the optimal DFCs are input into the SVMs to detect faults and sequentially identify fault types from rotating machinery. The proposed method has been applied to detect structural faults from rotating machinery, and the efficiency of the method has been verified using practical examples.