摘要： 针对标准粒子群优化(PSO)算法存在收敛速度慢、容易陷入局部最优的问题，提出一个改进的PSO算法，该算法设计一种新的惯性权重，在粒子搜索的不同阶段采用不同的计算公式计算惯性权重，并引入自适应变异策略和线性变化的学习因子。实验结果表明，该算法的收敛性等性能比基本粒子群算法有明显提高，能较好地解决非线性问题。

关键词: 粒子群优化, 惯性权重, 自适应变异, 服务组合优化

Abstract: As the Particle Swarm Optimization(PSO) algorithm has some shortcomings of slow convergence and easy to fall into the local extreme value, this paper presents a improved particle swarm optimization with a new inertia weight. In different stages of the algorithm run, a corresponding formula is used to calculate the inertia weight. In Addition, adaptive mutation and linear-changed learning factor are introduced. The relational test simulation experiment is carried out. Experimental results show that the improved algorithm is feasible and efficient, it can solve norlinear problem.

Key words: Particle Swarm Optimization(PSO), inertia weight, adaptive mutation, service composition optimization

中图分类号: 

• TP 301.6