Aiming at the problem that it is difficult to denoise the telemetry vibration signal with rich frequency content in aircraft test, a denoising method of telemetry vibration signal based on empirical mode decomposition (EMD) and time-frequency analysis is proposed, and the noisy signal is decomposed by multi-scale EMD. The intrinsic mode function (IMF) at different scales are obtained, and the corresponding time-frequency representation is obtained by short-time Fourier transform (STFT), and the filtering threshold is set according to the time-frequency energy distribution. Filtering is performed in the time-frequency domain. And performing STFT inverse transformation on the filtered time-frequency representation, and accumulating and summing to obtain a denoised signal. The method has better noise suppression capability and protects the detail information in the signal from being lost. Computer simulation and real data processing results show that compared with EMD denoising and wavelet denoising, this method improves the SNR by 2.8 and 2.4 dB in the simulation condition, and improves the SNR by 14.3 and 10.1 dB in the real signal, which proves that this method has practical application value in the telemetry vibration signal preprocessing.