In order to improve the analysis accuracy of loosening blasting vibration signals,a hybrid denoising method based on local mean decomposition(LMD),multiscale fuzzy entropy(MFE),and singular value filtering(SVD) was established.Firstly,the vibration signal was decomposed by LMD method to obtain a series of product components(PF).Then,the blasting vibration signal was preliminarily denoised by calculating MFE and correlation coefficients.Finally,the real signal components were denoised and extracted by SVD filtering on the residual noise of the main PF components.The results show that the proposed LMD-MFE-SVD denoising method can effectively deal with the noisy PF components.For the simulated signal with multiple components with noise,the LMD algorithm is more efficient than the EMD algorithm.Furthermore,the signal-to-noise ratio(SNR),root mean square error(RMSE) and percentage of distortion(PRD) of the proposed LMD-MFE-SVD method are significantly improved by 11.73%,22.07% and 9.25%,respectively,compared with the LMD algorithm,which indicates that the noise reduction efficiency is considerable.According to the waveform and spectrum comparison of the measured loosening blasting vibration signal after denoising by the proposed LMD-MFE-SVD method,the denoised signal waveform is more concentrated with most of the signal information retained.The frequency spectrum is clearer,and the signal frequency peaks are effectively displayed.