摘要
采用粒子群算法处理约束优化问题时,由于约束条件使得解空间成为非凸集合,粒子容易陷入局部最优,因此在搜索过程的不同阶段,提出变步长因子的粒子群算法,实验证明改进的算法是可行的,且在精度与稳定性上明显优于采用罚函数的粒子群算法和遗传算法等其它一些算法.
In this paper, the particle swarm optimaziton handles are used to deal with constraint optimal problems. Owing to constraint conditions, searching space is not bulgy and particles are easy to be limited to local optimal. Therefore, we advance to the PSO of searching different scale gene in different phases during the searching process. Nttmefical results show that the improved PSO is feasible and can get more precise results than particle swarm optimization by using penalty functions and genetic algorithm and other op optimization algorithms.
出处
《佳木斯大学学报(自然科学版)》
CAS
2006年第3期340-342,共3页
Journal of Jiamusi University:Natural Science Edition
关键词
粒子群算法
动态罚函数
变步长因子
particle swarm algorithm
dynamic penalty function
scale gene