Fault Feature Extraction for Shaft System of Hydraulic Machinery Based on Wavelet Packet

Abstract

Shaft fault is the most common fault for hydraulic machinery. In this paper, wavelet packet energy spectrum analysis method was used for multi frequency bands division of shaft monitoring signals. The variable bands frequency energy can construct feature vectors needed for fault diagnosis based on support vector machine, by integrating the procedure of wavelet packet analysis method and online monitoring technology, feature can be extracted in real time, and it make possible for real-time fault diagnosis and prediction of hydraulic machine.