差分形式粒子群优化算法  

Particle Swarm Optimization with Difference Form

作　　者：袁代林[1] 

机构地区：[1]西南交通大学数学学院,四川成都610031

出　　处：《计算机仿真》2015年第8期294-297,共4页Computer Simulation

基　　金：中央高校基本科研业务费专项资金(2682013CX037)

摘　　要：针对粒子群优化算法(PSO)理论研究上的困难和容易陷入局部最优的问题,分析了以位置迭代方程表示的差分形式PSO,以便于从理论上分析PSO以及开发其它具有良好优化性能的PSO算法。指出了位置迭代方程并不是真正的差分方程,只是将其视为差分的形式。利用差分形式PSO容易开发出其它形式的PSO算法。通过探索、改变位置迭代方程中各项的表达形式,获得了三种新形式的PSO算法。函数优化算例的结果说明三个新PSO算法在平均最优函数值和收敛率方面与原始PSO相比较均有不错的表现。新形式PSO的研究为开发其它具有优良优化性能的PSO算法提供了思路。It is necessary to overcome the problems about particle swarm optimization （PSO）, such as the diffi- culty for theoretical research and falling in easily local optimum. The PSO with difference form was analysed, which was expressed by position iterative equation. So it is easy to analyse PSO in theory and exploit other PSO algorithms possessing good optimal performance. It was pointed out that the position iterative equation is not a true difference e- quation and only is treated by difference form. Other PSO algorithms can be exploited easily by the difference form. Three new PSO algorithms were obtained by exploring and changing every term expression in position iterative equa- tion. Through the examples of function optimization, three new PSO algorithms show preferable optimization abilities comparing with the primitive PSO in average optimal function value and rate of convergence. The research of new PSO algorithms provides idea for exploiting other PSO algorithms possessing excellent optimal performance.

关 键 词：粒子群算法 差分形式 最优化 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]