基于信息微传递机制的粒子群算法  

Particle swarm optimization based on information micro-transmission mechanism

作　　者：闫李 李国森 瞿博阳[1] 马佳慧 YAN Li;LI Guo-sen;QU Bo-yang;MA Jia-hui(School of Electronic and Information Engineering,Zhongyuan University of Technology,Zhengzhou 450007,China)

机构地区：[1]中原工学院电子信息学院,河南郑州450007

出　　处：《计算机工程与设计》2021年第11期3135-3141,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(61673404、61876169、61873292、61976237);河南省高等学校重点科研基金项目(19A120014、20A120013);河南省科技攻关基金项目(182102210128);中原工学院青年骨干教师基金项目(2018XQG09、2019XQG03)。

摘　　要：针对粒子群算法易陷入局部最优,导致收敛速度慢、寻优精度不高的不足,提出基于信息微传递机制的粒子群算法(IMPSO)。引入信息微传递机制,将整个种群划分为多组多层,每组粒子逐层学习最优信息,防止算法早熟;采用逃离策略,当检测到粒子具有趋同行为时,改变粒子的飞行方向,增强算法寻优能力;使用动态边界化策略,动态缩小粒子的寻优区域,提高算法搜索效率。实验结果表明,IMPSO算法在收敛精度、收敛速度方面优于其它6种算法。Particle swarm optimization(PSO)algorithm is easy to fall into the local optima,which leads to low convergence speed and low optimization accuracy.A particle swarm optimization algorithm based on information micro-transmission mechanism(IMPSO)was proposed.An information micro-transmission mechanism was introduced.The whole population was divided into multiple groups and multiple layers.Each group of particles learned the information of the optimal solution layer by layer to prevent premature convergence.An escape strategy was adopted to change the flying direction of the particles when the convergence behavior was detected,which enhanced the search ability of the proposed algorithm.A dynamic boundary strategy was utilized.This strategy reduced the search space of the population dynamically,which improved the search efficiency of the algorithm.Experimental results show that the proposed IMPSO algorithm is superior to the other six algorithms in terms of the convergence precision and the convergence rate.

关 键 词：粒子群优化 寻优精度 寻优性能 收敛速率 函数优化 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]