Abstract: In the present paper, the measured signal of vibration system is denoised based on Wavelet Packet analysis, which has the properties of both time and frequency analysis. In the denoising process, the compromised threshold wavelet denoising method is adopted, which has the advantage of both the soft threshold and the hard threshold of the wavelet denoising method. Due to the compromised factor, flexible handling in analyzing the signals is permitted. By using timeenergy Wavelet Spectrum analysis to the denoised signal, the principal component is clearly shown. Based on the calculated energy distribution on the frequency domain, the characters of the signal are displayed clearly, which provides a theoretical method for fault diagnosis and fault identification of mechanical vibration system. Thus, the developed analytical method in the present paper has broad application in the fault diagnosis and fault identification of mechanical vibration system and its components.