Research On Blind Source Separation Problem Based On NNGA

Blind Source Separation(BSS)is a traditional method of signal processing,it is intended to separate and recover the source signals only by observed mixed signals but otherwise unobserved the source signals and the channel.It has very important theoretical significance and practical value and has made outstanding contributions in the field of speech signal processing,Biomedical engineering,image processing,and so on.Firstly,this paper introduces the problem of blind source separation and the significance of the research;analyzes the domestic and international development;and describes the relevant theoretical knowledge of blind source separation,including the mathematical model of blind source separation,the feasibility of signal separation,the uncertainty of separation results,the independence criterion and estimate method of blind source separation,the general solving process of blind source separation;then discusses and points out the deficiency of natural gradient algorithm,and introduces the algorithm of blind source separation in this paper-nonholonomic natural gradient algorithm.Then,in order to solve the contradiction between the convergence speed and steady-state error in fixed step algorithms,this paper proposes to use separating degree to control the step-size of nonholonomic natural algorithm.Because the variability of the new algorithms step-size is based on separating degree,its learning ratio is chosen adaptively according to separating degree,therefore it can achieve better separation effect.On the other hand,it is hard to get the balance between divergence speed and steady-state performance using only some specific estimate function in an algorithm,therefore,this paper puts forward the concept of optimal selective function,namely,selecting different estimate function in different stage to increase the flexibility of the function,furthermore,a variable step-size which is based on the gradient of cost function is also applied to the proposed algorithm.Computer simulation results show that compared to the usual algorithm,the two approved algorithms are more feasibility and superiority and can achieve better separation effect.Finally,the work of this paper is summarized,the existing problems and the future development direction in the research is prospected.