一种单个粒子自适应修正的粒子群算法

在各类优化问题的解决过程中,群智能优化算法的局部搜索与全局搜索性能都起着重要的作用。在粒子群优化算法中,惯性权值的引入对粒子群算法的收敛性与稳定性都具有一定的影响。因此,在分析现有权值递减策略的基础上,提出一种基于单个粒子适应值的权值修正策略,区别对待同次迭代中适应值好与差的粒子,通过不同的权值赋值策略,以充分发挥各粒子的优势,以增强全局搜索和跳出局部最优的能力。通过对标准测试函数所做的对比实验,该策略可以使粒子在搜索初期获得更好的多样性,使粒子具有更强的摆脱陷入局部极值点的能力;在搜索末期可以加快粒子收敛速度以提高粒子群优化算法的快速性能。改进算法有效减少了早熟的发生,提高了粒子的收敛性能,取得了比较满意的仿真结果。

Particle Swarm Optimization Algorithm with Self Adapting Inertia Weight

In the process of solving all kinds of optimization problems,local searching and global searching performance of swarm optimization algorithm play an important role.In particle swarm optimization(PSO) algorithm,the inertia weight has a certain effect on convergence and stability.Inspired by the effect of inertia weight on convergence of PSO,a new modified strategy for inertia weight is proposed based on fitness value.Comparative experiments of benchmark functions indicate that this new strategy could make the particles various to get the strong ability to keep from plunging local optimum and improve the astringency speed in the end of searching.Experiment results show that it is effective for prematurity and improve the ability of convergence.