Research On Blind Source Separation Problem Based On Swarm Intelligence Algorithms

Blind signal separation technology is one kind of method for recovering source signal from observation signal and it has broad application prospect and development potential in various fields such as speech, image, communication and biomedicine. Research on blind signal separation technology has been a hotspot in signal processing and intelligence computation. Swarm intelligence optimization algorithm is a kind of optimization algorithm using the principle of survival behavior of organism in nature and it is an efficient method for solving complex optimization problem. Therefore, there is fine development prospect for using swarm intelligence optimization algorithm in blind signal separation.Linear mixture blind signal separation was researched on the basis of study on swarm intelligence optimization algorithm and principle of blind signal separation in this paper. The main work can be expressed as follows:(1) To solve the problem that the original particle swarm algorithm is easy to fall into the local optimum, an adaptive particle swarm optimization algorithm based on mutation operator was proposed. The absolute value of Negentropy was used as objective function in the algorithm and improved particle swarm optimization algorithm was used for optimizing it, Source signal can be separated efficiently. Through simulation for separation on sub Gaussian signal and mixture of super-Gaussian signal and sub-Gaussian signal, similarity coefficient and Signal to Noise Ratio of two performance evaluation criteria compared with basic particle swarm algorithm, the result demonstrate the validity of the new method.(2) To overcome the shortcoming of the original imperialist competitive algorithm is easy to fall into local optimum, an imperialist competitive algorithm optimization algorithm based on Chaos theory was proposed. The value of mutual information was used as objective function in the algorithm and improved imperialist competitive algorithm was used for optimizing it. Through similarity coefficient and Signal to Noise Ratio of two performance evaluation criteria compared with the original algorithm, test result shows the better global searching ability.