Abstract:It is difficult to analyze the noise-containing aliasing signal with the traditional signal processing method,which also brings difficulties to the monitoring of the operation condition and fault diagnosis of the rotating machinery.Therefore,combined with the characteristics of strong periodic of rotor vibration signal,a sparse decomposition method of cross correlation was proposed.The simulation signal was compressed and sensed,so the dimension reduction of the signal was realized.The discrete Fourier dictionary was constructed,the sparse representation of the signal was obtained by orthogonal matching tracking algorithm.The correlation between the reconstructed signal and the original signal was calculated by using Pearson coefficient,and the most appropriate sparsity K was selected to reconstruct the noise reduction of the simulation signal.The sparse coefficient was multiplied with the atoms in the dictionary,and different frequency components in the simulation signal were separated; the vibration signal of rotor was analyzed,and the rotor frequency and its frequency doubling components were successfully extracted.The results show that compared with the traditional signal processing method,the noise-aliasing signals processed by using the proposed method are easier to analyze,which is helpful to the running condition monitoring and fault diagnosis of rotating machinery.