A modified threshold function for de-noising of rolling bearing vibration signal

To overcome the shortcomings of traditional threshold de-noising methods, a modified threshold function is proposed for wavelet packet threshold de-noising of rolling bearing vibration signal. The structure of the proposed function is simple without introducing adjustment parameters. The new method not only circumvents the oscillation caused by the discontinuity of the traditional hard threshold function, but also eliminates the inherent deviation resulting from the traditional soft threshold function. Based on the simulation comparison, the proposed method reduces effectively the noise with high signal-to-noise ratio and low root mean square error. The experimental result demonstrates that the amplitude of fault characteristic frequency is kept very well with low noise, implying the effectiveness of the proposed method. The presented method, with simplicity and effectiveness, is capable of providing a promising candidate for fault diagnosis of the rolling bearing and the other mechanical components.