The Improved Particle Swarm Optimization Algorithm And Application Research

Since the 20th century, the optimization domain has developed quickly due to the birth of swarm intelligence. The scientists obtained the enlightenment in the research of the biology swarm behavior and proposed many algorithms including PSO algorithm on the basis of the swarm intelligence. PSO algorithm is a novel army in intelligent optimization algorithm. It is an adapted random algorithm based on the population search strategy and gets a lot of concern. This algorithm is simple and the convergent speed is quick, in addition, it has few parameters and easily programming. In the project practice, PSO has already been widely applied in the function optimization, the parameter optimization, neural network training and PID parameters tuning.This thesis begins from the basic principle, parameter choice and application of PSO and presents a new PSO algorithm in which the existing problem of traditional PSO algorithm can be partly avoided, including slow convergence and getting into local optimum of traditional PSO algorithm. In short, the idea is that the parameters of traditional PSO algorithm can be adjusted to make the variation on speed and aspect of process in particle optimization. On that way, variety of population is increased and the searching ability is enhanced. Schaffer's testing functions are used to simulation, and the testing result proved that improved PSO is feasible and effectual.At the first part, the present situation in PSO algorithm is introduced. Secondly, the theory and application of PSO is summarized. The novel algorithm is presented and plunged into the function test in the third part. Last part is the application of the algorithm to the multi-objective optimization research.