The Research Of Blind Source Separation Algorithm Based On RBF And Cloud Adaptive Particle Swarm Optimization

Blind source separation is a signal processing method, which was developed in thelate1980of the20th century，It was produced by neural network and informationtheory，Its essence is to aim at non-Gaussian source signals. Under the conditions ofassuming statistical independence，to recovery original signals from the observedsignal. Because of a little of prior knowledge, it was used widely. At the present time,much of work has to be carried, such as in earthquake，wireless telecommunications，Image signal processing and signal analysis and other fields.The theories of this article based on blind source separation，References to achieveblind source separation algorithms，For example, natural gradient algorithm，geneticalgorithm，Particle Swarm Optimization algorithm and so on. But it has some defect,global convergence properties of unsatisfactory，Convergence is slow，Depends on alinear function of selected，and Evaluate complex defects. In order to improve thedefects，This paper studies a blind source separation algorithm based on radial basisfunction neural network，Greatly reduced algorithm of computational complexity，Than traditional gradient algorithms of experimental effect.Particle Swarm Optimization algorithm in optimal has a strong advantage，It hassimple principle，Less set parameters，Facilitate the achievement of the advantages。Put it with the most important advantage is that no blind source separation withnonlinear function select. But there are drawbacks，Blind source separation of datamore，Particle Swarm Optimization algorithm for premature，In local optimization. Inorder to overcome this disadvantage of Particle Swarm Optimization algorithm，Blindsource separation based on Particle Swarm Optimization algorithm in this article hasbeen improved. Combining cloud theory，According to the tendency of randomnessand stability characteristics of cloud droplets，Studying a cloud-based theory ofAdaptive Particle Swarm algorithm for blind source separation。Improved algorithmto effectively avoid premature convergence，Improved global search capabilities andconvergence rate. And it well had done to contain noise in mixed-signal separationtasks.