Abstract:In the study of feature extraction and classification of microseismic events for real-time on-line monitoring of underground mines, the recognition efficiency often depends on the quality of training samples and test samples. In order to improve the quality of data samples and remove the noise doped in signals, the ensemble empirical mode decomposition (EEMD) method was used to pre-process the microseismic signals of underground mines. A series of intrinsic mode functions (IMF) and a residual component spread from high frequency to low frequency were obtained by using EEMD analysis method. The noise and residual components in IMF were removed by calculating the energy proportion of each component, and then the residual components containing the main information of the mine microseismic signal were reconstructed to obtain the microseismic signal after denoising. Through signal simulation experiment and case analysis, the wavelet pretreatment method was compared. The results indicated that the method not only overcomed the disadvantages of wavelet threshold and decomposition function selection, but also significantly improved signal-to-noise ratio, retained signal shape and achieved ideal denoising effect by using EEMD adaptive decomposition characteristics.