Fast Independent Component Analysis Technique Based on Match Pursuit and Its Application to Extraction of Bearing Composite Fault

It is very difficult to extract and separate bearing composite fault information submerged in strong noise environment; therefore,the fast independent component technique based on match pursuit is proposed. Based on the match pursuit of Gabor atom,this method can first improve the signal to noise of single-channel mechanical signal and make the signal dimension raise,obtain many reconstruction signals,which make up multi-dimension signal and let signal dimension-raising come true. The denoising composite fault signals are separated by the fast independent component analysis technique. The analysis result of roller bearing composite fault experimental data shows that this separation of composite fault signals in strong noise environment has the feasibility and validity.