The Improved Particle Swarm Optimization Algorithms: APSO And DPSO

Particle swarm optimization(PSO)is a swarm intelligence technique developed by Eberhart and Kennedy in 1995.PSO has made considerable progress and lead to numerous applications in various fields(e.g.neural network training and engineering optimization).In PSO,each member of the swarm studies the informations from itself and the other members to do the new move.As one particle follows two targets to search,one is previous position of the particle and the other is the best position of swarm(the Gbest model)or the best position of neighborhood(the Lbest model).PSO is simple,robust and efficient. However,PSO has the disadvantage of easily trapping into local optimum on solving multi-modal functions,and poor local search in the later stage.How to solve the above problems and improve performance of PSO? This is an open question to the study of PSO.The purpose of the study has two handles,(1)to study a good PSO algorithm in theory, and(2)to using this algorithm to solve not only benchmark functions but also packing problems.The ultimate aim is to take it as an effective part of hybird method used for the layout design of satelite module.According to the study of particles' trajectories,new velocity updating formulas is designed,furthermore,two improved PSO algorithm which are Active Target Particle Swarm Optimization(APSO)and Detecting Particle Swarm Optimization (DPSO)are proposed.The main contributions are as follows:(1)An Active Target Particle Swarm Optimization(APSO)is presented.APSO uses new three-targets velocity updating formula,i.e.,the best previous position,the global best position and a new target position(called active target).APSO has the advantages in good ability of jumping out the local optimum and the ray seach ability of complex method; however,it has the disadvantages in adding some extra computation expenses.(2)A Detecting Particle Swarm Optimization(DPSO)is presented.In DPSO,several detecting particles are randomly selected from the population and the detecting particles use the newly proposed velocity formula to search in spiral trajectories.As a whole,the detecting particles and common particles would do the high performance search.DPSO tries to improve PSO's performance on swarm diversity,the ability of quick convergence and jumping out the local optimum.However,it also has the disadvantages in adding some extra computation expenses as APSO.The experimental results from several benchmark functions demonstrate good performance of APSO and DPSO.The experimental results from packing problem and the layout design of satelite module problem verify of the feasibility and validity of APSO and DPSO,and demonstrate APSO and DPSO to push forward the theory study of improved method.