SVD曲率谱降噪和快速谱峭度的滚动轴承微弱故障特征提取

iming at the low signal to noise ratio, weak fault information in bearing vibration, and fast kurtogram failed to select resonance center frequency and filter bandwidth accurately, an adaptive envelope method based on SVD curvature spectrum noise reduction and fast kurtogram is proposed. Fast kurtogram is able to automatically determine the band-pass filter's center frequency and bandwidth by the maximum kurtosis and thus calculate the Hilbert envelope. Associated with the SVD curvature spectrum noise reduction, the vibration signal is firstly decomposed by SVD, through singular values's curvature spectrum to reduce noise. Then use fast kurtogram to calculate kurtosis, and Hilbert envelope determined by the maximum kurtosis. Through the actual signal in rolling bearing shows that the method has a good effect.