Application Research On Intelligent Algorithm In Signal Blind Source Separation Technology

Blind source separation technology refers to the signal source and the transmission channel completely or partially unknown circumstances, the signal separation and extraction process using only the multi-channel sensor or antenna array of observed signals to achieve source signal. In recent decades, blind source separation technology has become a hot spot in the field of signal processing, etc, signal processing and has achieved many significant results. At present, the blind source separation technology has been widely applied in speech signal separation and identification, image processing and recognition, biological signal processing and machine fault diagnosis, and other fields.This paper first introduces the background and research significance of blind source separation technology, as well as domestic and foreign research situation of the blind source separation, and the main research direction; then the instantaneous linear mixed models were studied in detail when discussing some of the basic theory of blind source separation, including constraints blind source separation, signal data center and whitened pretreatment technology. Commonly used objective function and blind source separation (maximum likelihood estimation and maximum entropy, etc.) and optimization algorithm (oil painting algorithm based on adaptive and fast independent component analysis algorithm, etc.); Then in the interference suppression of wireless voice transmission system, ICA blind source separation algorithm is proposed to improve the sliding window; From the speech signal blind source separation problem, introducing bionic intelligent algorithm instead of traditional optimization approach; Then introduces several common bionic intelligent algorithms, such as genetic algorithm (GA) and particle swarm optimization (PSO), flora algorithm (BFO) and its improved algorithm, and applies them to the voice of the mixed signal blind source separation.In this paper, the short-term stability problem of wireless voice signal transmission channel, gives the improved algorithm based on sliding ICA, simplifies the traditional algorithm, to a certain extent, reduce the computational complexity. For bionic intelligent algorithm in the early stage of the optimization need strong global search ability to achieve optimal solution area as soon as possible, and later to the local search ability in local optimization to improve search precision demand, this paper introduces the influence of iterative algebra to general adaptive particle swarm optimization (APSO), meet the demand of the algorithm of weight before and after. Studied the quantum particle swarm optimization (QPSO) algorithm and adaptive flora (ABFO), through simulation experiment, proves that the algorithm in the time-varying system and time invariant system signal blind source separation in all has faster convergence speed and accuracy.