Abstract: Acoustic emission signals resulting from rock fracture contain time-frequency information. Exploring the relationship between signal parameters and precursory characteristics of rock fracture has practical significance for the prediction of rock engineering disasters. The traditional Pseudo Wigner-Ville Distribution (PWVD) algorithm can highlight the local characteristics of the signal by windowing the signal,but the signals collected in the test are generally real values. If PWVD is directly used for time-frequency analysis,there may be a false spectrum. To solve this problem,a time-frequency analysis method of PWVD based on the Hilbert transform(HPWVD) is proposed to study the time-frequency of acoustic emission signals. Firstly,the acoustic emission signals of the tungsten rock and sandstone are collected during uniaxial compression tests. Then the parameter analysis method is used to analyze the cumulative ringing count and the distribution characteristics of cumulative energy of the two kinds of rock samples,so as to divide the fracture stage and determine the critical point of the fracture stage. Finally,the HPWVD algorithm is used to obtain the time-frequency characteristics of acoustic emission signals before and after the critical point.The results show that using the HPWVD algorithm to analyze the signal with time-frequency can eliminate the phenomenon of the false frequency spectrum,and can accurately obtain the time-frequency information before and after the critical point,that is,the energy of the rock before fracture will be redistributed in the new frequency band,provideing a basis for rock fracture prediction.