Single Channel Blind Source Separation Algorithm For Stratum Elastic Wave Channel

This subject comes from National Natural Science Foundation of China's“Continuous seismic source-based ground reflection type slot wave fine detection theory research".When the source signals and its channel in stratum are unknown,only one detector at the receiving end receives the observation signal,and the source signals is separated from the single observation signal.This subject aims to design a single-channel blind source separation algorithm suitable for stratum elastic wave channel.Based on the analysis of the transmission characteristics of elastic wave signal in the stratum,it is clear that the design of blind source separation algorithm for stratum elastic wave channel needs to overcome the difficulties of single observation signal and serious attenuation of source signal.A virtual multi-channel blind source separation algorithm based on ensemble empirical mode decomposition-principal component analysis-fast independent component analysis(EEMD-PCA-FastICA)is improved.The improved algorithm is applied to the single channel blind source separation of the stratum elastic wave channel,the simulated simulation experiments are performed using MATLAB.By comparing and analyzing the simulation results,it is found that there is a disadvantage that the improved algorithm does not have obvioustime-consuming improvement and poor separation effect when processing non-periodic signals.Aiming at the shortcomings of the improved algorithm,a single-channel blind source separation algorithm based on EEMD--T-PCA--PA-FastICA is proposed,the innovation of the algorithm mainly includes the following two aspects:1)two principal component analysis is proposed,which reduces the number of Intrinsic Mode Function component,thereby reducing the number of iterations of the algorithm and shortening the consumption of the algorithm;2)a fast independent component analysis algorithm based on Pade approximation is proposed to make up for the shortage of algorithm's convergence speed and degraded signal separation performance when there are too many intrinsic mode function components.Aiming at periodic and non-periodic signals,the single-channel blind source separation simulation experiment for simulating stratum elastic wave channel using EEMD-PCA-FastICA improved algorithm and EEMD--T-PCA--PA-FastICA algorithm respectively;by comparing the results of the experiment,it is found that using the EEMD--T-PCA--PA-FastICA algorithm has the advantages of high separation signal fidelity,fewer iterations of the algorithm,and short time-consuming advantages,especially in the separation of non-periodic signals.