The Research Of Blind Source Separation Based On Improved Particle Swarm Optimization

Blind source separation (BSS) is to separate the source signal from the received signals without any prior knowledge of the source signal and the transmission condition. As a new technique in digital signal processing field, BSS has very important theoretic meaning and practical value. BSS has been widely used in audio signal processing, wireless communication, noise elimination, biomedicine signal processing, earthquake signal processing, image signal processing and other fields.The current blind source separation has some problems. The paper applies the particle swarm optimization and its improvement to blind source separation. The main works of the paper can be summarized as follows:(1) Summarizing the origin of BSS and the research progress at home and abroad, the paper introduces the fundamental principle of BSS and simply summarizes the current algorithms. It also analyzes the basic postulates of BSS problems and typical evaluating index of BSS algorithm performance.(2) Aiming at the questions of low convergence rate of BSS based on particle swarm optimization (PSO), the paper applies the PSO based on simulated annealing method. After comparing the basic PSO algorithm and some improved algorithms, the paper selects the simulated annealing PSO which has the fastest convergence rate as the optimization method. Then the algorithm is applied to BSS and the paper designs the basic steps of the algorithm and makes some relevant simulation experiments which prove the algorithm is valid.(3) PSO-based BSS is apt to getting into local most optimization problems. The paper applies the PSO with grads acceleration to BSS and designs the basic steps of the algorithm and does some relevant simulation experiments. The experiment results reveal that applying the PSO with grads acceleration to BSS can make the movement of particle more pertinent and the movement is more efficient. It not only improves the convergence speed of PSO algorithm, but also makes the improved PSO algorithm is more efficient it the aspect of global convergence, stability, etc. So it improves the veracity and efficiency of the algorithm.(4) After summarizing BSS algorithm of convolved mixture and its applications, the paper applies the improved PSO algorithm to convolved mixture BSS and proposes the PSO algorithm based on simulated annealing method. The paper also does some experiments of BSS based on PSO algorithm with grads acceleration and analyzes the simulation results.