In order to solve the problem of frequency confusion and error frequency in the process of analysis,the spectrum analysis algorithm of wavelet transform signal was improved.The correctness and feasibility of the improved algorithm was verified by the way of selfdefined combination function.The improved algorithm of signal processing was applied to the processing of turning surface topography signals.The frequency domain was divided into complex signals,and the spectrum analysis of each frequency band reconstructed signal was carried out.Finally,the signal extraction and characterization of each frequency band were realized.Combined with the influencing factors of the surface roughness of turning,the corresponding relationship between the signal of each frequency band and the influencing factors was obtained,which provided theoretical support for improving the roughness of the turning surface.