The Research On Particle Swarm Optimization Algorithm With Dynamic Topology

The optimization problem together artificial intelligence and computer science with mathematics science, which have made wide using in the industrial and economic management. Linear and non-linear are the two basic optimization problem and non-linear problem is difficult to solving in the limited time and customer requirement. It could be used in the optimization management of resource, the scheme to optimize designs, and income maximized of the enterprise.Particle swarm optimization as a basic swarm intelligence algorithm has self-organizing, collaborative behavior and information interaction. Through the nature of the conduct, PSO is one of the most widely used methods of simple algorithm. And have to rely on the question of the initial conditions. The solution of the process can be better integrated approach to the overall performance. Evolutionary strategies and topology structure is two important branch of the PSO research. Topology structure can give us the interaction information of the particles. Good topology gets better result. Dynamic topology is that the topology structure changing with evolutionary processing.We have analyzed the topology of the PSO and began with the interactivity level and then propose the sub-swarm topology. It can make the scope of the sub-swarm changing with evolutionary processing by convergence rate and sub-swarm rejection mechanism. The paper has done some work in the behind:1. By analyzing swarm intelligence, we show the basic process of the PSO.2. Analyzed the topology structure of the PSO and proposed an information interaction level. And then proposed the tow-layer alterable sub swarm PSO.3. Using five popular benchmark functions to test the proposed algorithm performance. And the result shows us that the algorithms have made good convergence rate and global searching.The research can show us that the two-layer alterable sub-swarm particle swarm optimization can balance the global optimization and the convergence speed. It also can get good progress of convergence of precision and robustness.