种群分类粒子群改进算法研究

针对粒子群算法在陷入局部最优时难于跳出的缺陷,提出一种改进的粒子群算法.该算法首先利用粒子适应值的统计规律对粒子进行分类,对属于不同类别的粒子采用不同的进化模型,对于利用完全模型进化的粒子,采用动态调整学习因子的方法,从而大大提高了算法的优化效率和优化精度.通过反复实验分析,得出学习因子随着进化推进的最优变化规律,并给出了学习因子的最佳函数表达式.仿真结果表明,利用改进的PSO算法优化4种具有代表性的基准函数,无论是在优化精度方面还是在优化效率方面,均较以往提出的PSO算法在性能上有本质的提高.

An improved particle swarm optimization algorithm based on population classification

Particle swarm optimization(PSO) algorithms have the disadvantage that once they find a local optimization it is hard for them to jump out and continue on to a global optimization.To solve this,an improved particle swarm optimization algorithm was developed in this paper.The algorithm uses statistical laws of particle fitness to classify particles,and takes different evolutionary models for different kinds of particles.For particles evolving in the full model,learning factors are adjusted dynamically,which can greatly enhance evolutionary efficiency and precision.Through a series of experimental analysis,the optimal changing rule of learning factor in the progress of evolvement was obtained.The optimal function expression of the learning factor were given.Simulation results showed that,compared with other PSO algorithms proposed before,it significantly improves both optimization precision and efficiency when the improved PSO algorithm is used to optimize 4 typical benchmarks.