约束松弛机制和反向搜索的粒子群算法

Particle swarm optimization based on relaxation mechanism for constrained conditions and opposition searching

提出了一种基于约束松弛机制和反向搜索的粒子群算法。约束松弛机制通过度量不可行解与可行解的距离,使边界位置的不可行解得以概率性存活,缓解了约束条件的不利影响;反向搜索机制则使不可行解迅速转向某一对称区域,提高了可行域的发掘效率。同时,利用局部最优位置的概念,使个体依据约束违反情况执行不同的搜索算子,克服了函数形貌的影响。通过六组标准约束优化问题的测试结果表明,所提出算法总体优于几种对比算法,最优解与理论值的相对误差由0~25.214%降低到了0~0.752%。

This paper proposed a particle swarm optimization based on relaxation mechanism and opposition searching. The relaxation mechanism treated constrained conditions by computing the distance between feasible and infeasible solutions, which made some infeasible solutions near the feasible region survive in a probability, and relieved the influences of constrained con- ditions to the optimization. The opposition searching guided infeasible solutions quickly to tend to symmetrical region so that the efficiency of finding feasible region was improved. Besides, the algorithm adopted different operators for different individu- als depending on the constrained violation to overcome the effect of the landscape, wherein the operators were designed based on the notion of local optimal position. Experiments on six benchmarks display that the proposed algorithm works better than other algorithms, and the relative errors between obtained value and theory value is reduced from 0 - 25. 214% to 0-- 0. 752%.