邻域搜索的粒子群优化算法及其性能分析

粒子群优化算法（PSO）是一种进化计算技术，是一种基于迭代的优化工具。但是，该算法的本身特性决定了算法不趋向于搜索接近极值点的解空间，造成了PSO算法最终解的局部极值性不好；并且，PSO算法需要充分的迭代才能够得到比较好的解，在迭代步数受到限制或者随时可能中途停机的情况下往往不能够得到比较好的解。根据PSO的这些不足，提出了邻域搜索的f-PSO算法，该算法在PSO的迭代步骤中每次更新全局最优解的同时采用一步局部寻优过程。实验表明，该算法具有很强的理论价值，在运算能力不足 、迭代不充分或中途停机的情况下，该算法仍然能够得到比较好的解。

A Local-Search-Based Particle Swarm Optimization Algorithm and Its Performance Analysis

Particle Swarm Optimization (PSO) is an evolutionary computation technique and an optimization tool based on iteration. However, the PSO algorithm does not search the solution space of the points closest to the extrema, which leads to the bad local extrema of the final solutions. Moreover the PSO algorithm needs enough iterations to get better solutions, and it usually cannot get better solutions with the limit of iteration steps or the situation of break at any time. Based on those shortages, the f-PSO algorithm which is based on neighborhood search is presented. The algorithm searches for local optimization solutions when it upgrades the global optimization solutions each time in the iteration steps of PSO.Experiments indicate that this algorithm has a strong theoretical value and good robustness, even with the lack of computation, insufficient iterations or break at any time.