The research of processing valve train dynamic data based on the wavelet packet analysis method

It´s often using digital filtering method to eliminate the interfering signal presently, but this method is difficult to determine the frequency range of interfering signal, it can´t distinguish interfering signal from real signal, and can´t totally eliminate the distinguish signal on the condition of not affecting real signal. According that different segment of every cycle dynamic data has different frequency response characteristics; this paper proposes a method using both FFT analysis of different segment and wavelet packet analysis together. First, do wavelet analysis of the dynamic data to eliminate the white noise as a pretreatment, then divide the dynamic data of one cycle into different segments basing on different segments having different frequency response characteristics, do FFT analysis of each segment to determine the frequency range of the interfering signal, use method of wavelet packet analysis and reconstruction to eliminate the interfering signal. This method shows a good result: Comparing with common using digital filtering method, this method improves 1.02 percent of the dynamic data accuracy, decrees 12 percent of valve speed distortion which is obtain from the integral of dynamic acceleration data, improve 7.71 percent of valve load speed accuracy. It provides new ideas to process valve train dynamic data.