The wavelet analysis can divide signal into different frequency bands and carry out signal decomposition and extracting the feature information of different parts of a machine. At the same time, FFT-FS spectrum zooming technology and AR spectrum estimate method can overcome the difficulty of the sampling frequency and sampling number so to get accurate frequency. As an example, vibration signal of reduction device is analyzed. As a result, the tooth faces of gear pair in the high-speed stage is deteriorated seriously and the bear in the high- speed shaft on the work side has fault feature.
展开▼