The Research Of Improved Particle Swarm Optimization Based On Multi-population

Particle Swarm Optimization is a random search algorithm based group collaboration by simulating the behavior of birds foraging. Because its simple concept, easy implementation, fast convergence, and it can effectively solve many practical problems, therefore, the algorithm proposed immediately has attracted wide attention from many scholars and gradually become a new hotspot.Although the particle swarm optimization algorithm is simple and has good results for the general optimization problem; but in the application of PSO algorithm for some complex, high-dimensional, multi-extreme optimization problem, the algorithm can not find the overall situation and can easily convergence in the local advantages. So, how to improve the global convergence performance of PSO algorithm is an important topic for PSO algorithm.In view of the deficiencies of particle swarm optimization algorithm's global convergence performance when optimizing the complex issues, this article identifies the shortcomings through analyzing the particle swarm optimization algorithm. Several improved particle swarm optimization algorithms are proposed and applied into math problems. The main works of the dissertation can be organized as follows:(1) In order to enhance diversity of species, particle swarm optimization algorithm of two-group exchange based on different evolutionary model and particle swarm optimization algorithm of three-group based on different evolutionary models are proposed. And give the corresponding algorithm, algorithm flow, and simulation results. The results show that: at optimization of complex functions, the new algorithms proposed to improve the global convergence properties are better than the standard PSO algorithm.(2)On the basis of the culture algorithm, a new collateral particle swarm optimization algorithm is proposed, which makes the particle swarm optimization bring into cultural algorithm frame. In the cultural algorithm frame, population space and belief space composed by particle swarm iterate dependently and collaterally, besides affect one another. Give cultural principle, algorithm flow, and simulation results about proposed algorithms. Compared to basic PSO, the improved parallel particle swarm optimization based on cultural algorithm has better global search capability. (3) In order to test the practicality of several new proposed improved optimization algorithms, several improvements proposed in this paper are applied to nonlinear equations problem, and get the solution of equations in the plural range of nonlinear equations. Compared to the traditional numerical methods, the algorithms don't need to require derivative information and the initial point of information of objective function, have no higher requirements on the nonlinear equations, and it is easy to come true.