Blind Source Separation Method of Dual-Channel Single Vibration Signal Based on Power Spectral Density

In this paper,a convolution-mixing model of vibration signal is established according to the actual vibration signals collected by the sensor. This model is simulated and separated in P-waves and S-waves by using the frequency domain convolution blind source separation algorithm based on power spectral density matrix and diagonalization, and separated the vertical and horizontal waves of the actual two-channel underground vibration signals, which lays a technical foundation for the study of seismic characteristics and the inversion of source location.