The Fault Feature Extraction Of Gear-box Based On Blind Signal Separation

Blind Signal Separation (BSS) can separate signals and its paraments from the original signals and the channel without prior information.So it can be used in fault diagnosis to separate mixed signal. In this paper,the method is used in the fault diagnosis of gear-box reciprocating compressor to exploit character parameter.Based on subject background,in this text,mainly do work on several parts below: 1.Analysed and explained the independent component analysis(ICA)theory and algorithm in detail;gave the class of core function and a way of its selection.FastICA algorithm is described in detail and simulated.The feasibility of the algorithm is verified through coherence determined by the separation of signals and simulation of signals.2.Study the mechanism of vibration of gear-box failure.Study typical failure mechanism of the gear and bearing vibration based on its common types and causes of failure. And fault feature extract in fault diagnosis.3.Simulated five kinds of typical faults of gear box and collected vibration signals under these conditions using laboratory test of gear-box. For multiple sets of experimental data, the first signal source separation, and then extract the data in time domain indexes after the separation as input of BP neural network. Diagnostic results show that the method proposed in this paper is feasible and can improve the accuracy and reliability of gearbox fault diagnosis.4.Developed the software platform for gear-box fault diagnosis based on blind source separation and feature extraction by matlab software. The platform uses parametric design,and realizes blind signal separation, feature extraction and fault diagnosis through the graphical user interface.