一种带随机选择机制的新型改进粒子群优化算法

标准粒子群优化算法对空间所有区域等概率搜索,降低了算法效率。借鉴遗传算法的思想,本文提出一种带随机选择机制的改进粒子群优化算法。该算法将适应值选择和粒子状态更新方程结合起来,通过赌轮算法选择机制使得粒子在适应值较小的区域尽可能的降低搜索概率,在最优解可能区域尽可能加大搜索强度,从而提高算法搜索效率。通过标准进化计算测试函数测试,实验结果表明对于复杂优化问题该算法优于标准粒子群优化算法和遗传算法。

An Enhanced Particle Swarm Optimization Algorithm with Stochastic Selection

In basic Particle Swarm Optimization （PSO） algorithm,particles search for optimal solution in problem space at the same probability,which leads to the low efficiency of PSO.Inspired by the idea of genetic algorithm （GA）,an enhanced PSO algorithm （EPSO） with stochastic selection is proposed.By combining particle status update equation with fitness selection,the proposed algorithm makes particles searching in different regions at different probability through roulette wheel selection and therefore the performance of PSO is improved.The proposed algorithm is demonstrated on standard evolutionary test functions and the results of tests show that it is superior to basic PSO and GA in the solution of complex optimization problems.