LOW FREQUENCY FAULT FEATURE EXTRACTION FOR GEARBOX BASED ON WAVELET TRANSFORM AND CONSTRAINED INDEPENDENT COMPONENT ANALYSIS

The effective low frequency fault feature hidden in gearbox measured signal is very weak for the influence of high frequency vibration and strong noise,and considering source noise,the extraction effect of low frequency fault feature with constrained independent component analysis（ c ICA） method directly is very poor. Aiming at this problem,a method of gearbox low frequency fault feature extraction based on wavelet transform and c ICA is proposed. This method can improve signal-to-noise ratio（ SNR） and nongaussianity of the analyzed signal,and enhance analysis effect of c ICA algorithm through wavelet multiresolution decomposition for the measured signal and reconstruction for some wavelet coefficients. But the analysis effect is not good for the c ICA method without WT denosing. Low frequency fault feature extraction experiments with a missing tooth and a chipped tooth were analyzed,and the results show that this method can effectively reduce the influence of high frequency vibration and source noise,and extract gearbox low frequency fault feature,especially weak low frequency fault feature. The proposed method provides an effective approach for low frequency fault feature extraction and fault diagnosis of gearbox