Blind Source Separation Method Of Elastic Wave Signal In Coal Measure Strata

This subject comes from the National Natural Science Fund Project"Precise Detection Theory of Reflective Groove in Coal Seam Based on Continuous Seismic Source".According to the National Natural Science Foundation of China, it is necessary to separate the source signals from the mixed signal by the theoretical model of the elastic wave signal and the source signal which can not be accurately known. The purpose of this study is to design a blind source separation algorithm for receiving elastic wave signals in coal bearing strata.(1)The development process of blind source separation is described in detail and the different applications of different algorithms are classifie-d,the basic theory and separation requirements of blind source separation are described,the noise environment and the transmission characteristics of elastic wave are determined, and makes clear the design requirements of the blind source separation algorithm of elastic wave signal in coal bearing strata.(2)The nonlinear function and orthogonalization formula of FastICA algorithm based on negative entropy are chose the right ,with the natural gradient algorithm are used to deal with the mixed elastic wave signal received by coal measure strata, and the mixed signal separation is realized successfully by Matlab simulation.It is verified that the non Gauss criterion and the likelihood criterion are essentially the same as the mutual information criterion,which proves that the FastICA algorithm based on negative entropy and natural gradient algorithm and the FastICA algorithm based on mutual information can achieve the coal formation receiving elastic wave signal based on blind source separation,independent component analysis algorithm has two inherent problems: separation order and separation signal amplitude are verified.(3)A new non orthogonal decomposition algorithm is proposed. The classical algorithm of independent component analysis to meet the needs of not less than the number of observation signal source signal,independent component must be the condition of non Gauss distribution of the premise,greatly reduces the practical analysis of elastic wave signal blind source separation in the coal bearing strata under independent component. The new non orthogonal decomposition algorithm does not need to meet the conditions in above, using the correlation analysis from a single observation signal in elementary function, and then, using these functions as a non orthogonal signal decomposition algorithm based on one mixed signals in each separate source signals of different. By using the algorithm of square wave, sine wave, attenuation wave modulation signal and random noise synthesis of single typical observation signals for simulation experiments show that the algorithm can not only from a single observation signal in the accurate extraction of the active signal, and compared the independent component analysis has recognized the separation sequence determination,separation of signal energy symbols the more excellent performance.