In the paper, pattern recognition methods for fault diagnosis of mechanical equipment such as bearing and gear were studied, the subjects include:1) Wavelet packet was used to extract the fault feature. Wavelet packet analysis is a method of the high-frequency part of decomposition. Referring to the more complex features of the vibration signal of the machinery equipments, the time domain character values were extracted from measured signal and Wavelet Packet coefficient variance for further analysis.2) Pattern recognition methods for fault identification were studied, the three pattern recognition analysis methods, Clustering, Neural Network and Fuzzy, were applied to fault pattern recognition based on the Wavelet Packet coefficient variance, Difference between varies pattern recognition methods was discussed.3) Combined pattern recognition methods the two methods, such as Fuzzy Clustering and BP NN, Fuzzy Inferencing and Adaptive Neural Network (ANN), was used to study the states of the objects. The analysis showed that better results can be achieved with combined method than with one pattern recognition mejtriock;... |