粒子群优化算法中惯性权值调整的一种新策略

惯性权值的设置对粒子群优化（PSO）算法的性能起着关键作用,现有的基于惯性权值的改进算法提高了算法的性能,但都把惯性权值作为全局参数,很难控制算法的搜索能力。本文在充分分析惯性权值的关键作用基础上给出一种新的惯性权值调整策略及其相应的粒子群优化算法,使用不同的惯性权值更新同一代种群。测试结果表明,新算法提高了算算法的性能,并具有更快的收敛速度和跳出局部最优的能力。

A New Strategy of Inertia Weight Adjustment for Particle Swarm Optimization

The setting of inertia weight plays a key role in the performance of PSO, so many improved PSO algorithms based on inertia weights are proposed. In these improved algorithms, the performance of the algorithm is really improved, but the same inertia weight is used to update the velocity of particles in the whole population and it is hard to control the search ability of PSO. A good investigation of the key role of the setting of the inertia weights is made and a new strategy of inertia weight a djustment and the corresponding PSO are proposed based on the investigation. In the new PSO, different inertia weights are used in updating the particle swarm in the same generation. The experimental results illustrate that the new PSO algorithm improves the performance of PSO,and it also speeds up the velocity of the PSO convergence and has the ability to escape from the local minimum.