Due to the influence of complex environment of tunnel blasting and the electromagnetic interference of instruments,the measured blasting vibration signals mostly contain high-frequency noise,which makes it ineffective to analyze related laws by the raw blasting vibration data.In order to obtain the real blasting vibration characteristics,a signal smoothing and noise reduction model based on the optimal variational mode decomposition(OVMD) and the multi-scale permutation entropy(MPE) is adopted,which is verified by the simulated superposition signals and measured signals.Firstly,the signal is decomposed by OVMD to obtain the band-limited intrinsic mode functions(BIMF).Then,the high-frequency BIMFs larger than the threshold set by MPE are removed as noise.Finally,the remaining BIMFs components are reconstructed to obtain the noise-reduced signal.The results show that the OVMD-MPE model can accurately identify the signal frequency information,and the first two order components can effectively reflect the effective contents of the superimposed signal,which is suitable for high-precision data analysis and feature extraction.Compared with EEMD-MPE and CEEMDAN-MPE models,the OVMD-MPE model has better noise reduction performance.The noise reduction error ratio,root mean square error and smoothness are increased by 22.05%,48% and 33.34%,respectively.The denoised curve is closer to the original signal and is more suitable for blasting signal analysis with different source distances.The blasting vibration signals measured during the construction of the right line of Shuangzishan Tunnel are concentrated in the middle and the low frequency bands below 200 Hz.The natural frequency of the lining structure is similar to the main frequency of the blasting signal,which means shock absorption measures need to be taken to ensure the construction safety of the tunnel project.