关键词: 小波分析, 独立分量分析, 降噪, 滚动轴承

Abstract: There is a problem of depending on experience of selecting parameters when wavelet analysis method is used to make noise reductions of vibration signals. And there is a problem of undetermined parameters when ICA method is used to make noise reduction of vibration signal. In order to avoid disadvantages of wavelet denoising and ICA method when they were used respectively. A method of wavelet denoising and FastICA independent component analysis based on negative entropy was proposed to deal with the noise signal of rolling bearings. Firstly, original signals were processed by wavelet denoising. Then the processing results were combined with the original signal to form input matrix of FastICA, and the FastICA noise reduction processes were performed. Finally, the effectiveness of the method was verified by vibration signal of rolling bearings. Experimental analysises show that this method improved the kurtosis of vibration signals and the purposes of noise reductions of vibrartion signals of rolling bearings was achieved.

Key words: wavelet analysis, independent component analisis(ICA), noise reduction, rolling bearing