摘要
为解决高维复杂CO问题,可将进化算法中保持物种多样性的思想引入基本PSO算法的方法。针对基本PSO算法在迭代后期粒子活性减弱的问题,引入‘吸引’和‘扩散’两个算子,对基本PSO算法的速度更新公式进行改进和考虑固定惩罚函数无法有效引导粒子向可行解方向搜索的缺点,提出LPFM方法替代固定惩罚函数法,以有效引导粒子进入可行解域,并在迭代后期加强对粒子的约束,使其不至因违背约束所获的收益大于所受的惩罚而收敛到不可行解域。最后对改进的PSO算法进行了试验,试验结果表明改进PSO算法对解决高维复杂CO问题是有效的。
By introducing the thought of keeping species diversity in Evolution Algorithm into basic Particle Swarm Optimization, an improved PSO algorithm was brought forward to solve multi - dimensional and complicated Constrained Optimization problem. In order to tackle the problem that the particles lacked activity in the anaphase of iteration, a velocity updating formula of basic PSO algorithm was improved by introducing the factors of ＇ attractive＇ and ＇ repulsive＇. In vies of the defect of PSO algorithm that it lacked the ability to lead particles toward the direction of feasible domain, the Stationary Punishment Function Method was proposed to replace Leading Punishment Function Method. At the end, the improved PSO algorithm was tested by two CO problems and the results proved that the im- provement of PSO algorithm was effective to solve multi - dimensional and complicated CO problems.
出处
《计算机仿真》
CSCD
北大核心
2009年第10期212-215,338,共5页
Computer Simulation
