To treat the tendency term and high-frequency noise of blast vibration signals,a signal processing method based on the CEEMDAN decomposition and the wavelet threshold has been introduced.First of all,a group of IMF components is obtained through CEEMDAN decomposition of radial blast vibration signals,and then FFT transformation is carried out for each component.The trend component is screened out and removed by spectral analysis.Meanwhile,noise-containing IMF components and pure IMF components are distinguished by cross-correlation function analysis of the remaining IMF components,which is further checked by autocorrelation function.Then,the IMF components containing noise are filtered by wavelet threshold.Finally,the IMF components and the pure IMF components after the wavelet threshold filtering are reconstructed as the processed signal.The results show that this method has obvious advantages in eliminating noise and trend term interference.Compared with the widely used processing methods(CEEMDAN method,wavelet threshold method,EMD-wavelet threshold method,EEMD-wavelet threshold method,CEEMD-wavelet threshold method),the signal-to-noise ratio(SNR) of CEEMDAN wavelet threshold method is 20.460,and the root mean square error(RMSE) is 0.086.The objective evaluation indexes of the method all show better treatment effect.The 3D time-frequency-energy diagram before and after processing reflects that the processed signal fully retains the real energy information in the 2~200 Hz frequency band,which further improves the accuracy of blasting vibration signal analysis and can be applied to the joint processing of trend term and noise similar to blasting vibration signal.