Research On Blasting Vibration Wave Propagation Characteristics And Blind Source Separation Technology

Based on the need of real-time monitoring system source location techniques for blasting mining activities in underground mines,this paper conducts in-depth research on elastic medium wave theory,viscoelastic media theory and convolutional blind source separation theory.Combining with the propagation characteristics of actual shock waves,a model for the propagation characteristics of blasting waves was established.Using a frequency-domain convolutional blind source separation algorithm based on the diagonalization of the power spectral density matrix to perform seismic phase extraction of the actual blasting vibration wave,the P-wave and S-wave separation is realized,and the relationship between the separation filter matrix and the hybrid filter matrix in the blind source separation algorithm was preliminary studied.It lays the technical foundation for the next focal position inversion and has high practical value and theoretical significance.Based on the propagation characteristics of blasting vibration waves:amplitude attenuation and frequency attenuation absorption characteristics.Amplitude attenuation is the result of both the wavefront diffusion and the absorption of the propagation medium,the shock wave propagation medium is closer to the viscoelastic medium.combined with the relationship between strain and displacement in the knowledge of seismic wave theory,the equation of motion balance produced by the normal stress of unit small volume element in uniform and slender rod medium,the constitutive equations of viscoelastic media models,respectively Maxwell Kelvin-Voigt and the standard linear viscoelasticity are shock wave propagation media,and three complete models of the propagation characteristics of the blast wave are established.An experimental scheme was designed to use the existing underground blasting vibration signal acquisition and data processing platforms to collect field data,by calculating the similarity coefficients between the three model prediction waveforms and the measured waveforms,the effects of the three propagation models are analyzed,and it can be concluded that the three propagation models can describe the propagation characteristics of blasting shock waves.Among them,the Kelvin-Voigt viscoelastic media propagation model has the best description of the propagation characteristics.According to the actual situation of the sensor to collect the blasting vibration signal,a two-channel convolution mixing model of single blasting vibration signal was established.Using a frequency-domain convolution blind source separation algorithm based on the diagonalization of the power spectral density matrix,the simulation separation of a single blasting vibration signal of two channels is realized,and the separation effect is verified by the signal-to-interference ratio and waveform similarity coefficient of the separated signal.For the first time,the above algorithm was used to realize the separation of P-wave and S-wave for a single actual vibration signal of two channels.A frequency-domain convolutional blind source separation algorithm with power spectrum density matrix combined diagonalization verifies that the separation filter matrix W(z)and the hybrid filter matrix A(z)satisfy the "ideal inverse relationship".