具有动态调节机制的多粒子群改进算法及应用  

Improved Algorithm of Multi-PSO with Dynamic Adjustment Mechanism and Its Application

作　　者：安丽霞[1] 张彩珍[1] 侯志伟[1] 包理群[2] 

机构地区：[1]兰州交通大学电子与信息工程学院,甘肃兰州730070 [2]兰州工业学院电子信息工程学院,甘肃兰州730050

出　　处：《兰州交通大学学报》2015年第1期71-76,共6页Journal of Lanzhou Jiaotong University

基　　金：甘肃省高等学校科研项目资助(2013A-127)

摘　　要：针对标准粒子群算法(standard particle swarm optimization,SPSO)的稳定性较差及易陷入局部收敛等缺陷,将粒子群体划分为多组粒子群,提出了一种子群粒子和其产生的精英粒子分两步协同进化的方案,采用混沌、高斯动态扰动粒子位置及云正态模型自适应动态调整惯性权重等动态调节机制优化粒子飞行轨迹,促进粒子又快又好的向群体最优目标飞行,以改善SPSO算法的全局寻优性能并提高多目标优化问题的多样性.采用新颖的误差适应度函数设计了FIR高通数字滤波器,并与基于RGA、PSO、CRPSO及典型Parks-McClellan算法的滤波器进行了对比与分析.仿真实验表明:基于具有动态调节机制的多粒子群改进算法及目标函数设计的滤波器,具有通带波动小,阻带衰减大的优势.Aiming at the limitation of standard particle swarm algorithm optimization（SPSO）in poor stability and being easy to fall into local convergence,a co-evolutionary scheme is proposed for sub-group particles and its group of elite particles respectively,in which the particles are divided into several groups.Dynamic adjustment mechanism is used to optimize the moving trajectories of particles,such as using chaos and Gauss in disturbance of particles＇ position and using the normal cloud model to adjust inertia weigh dynamically,etc.Particles can fly to optimal goal of the group quickly and well,improve SPSO algorithm of global optimization performance and increase the diversity of the multi-objective optimization problem.The algorithm is used to solve the coefficient combinatorial optimization problems in designing process of FIR digital filter.FIR highpass digital filter is designed by using a novel error fitness function.It is also compared with digital filers based on RGA,PSO,CRPSO and the typical Parks-McClellan algorithm.Simulation results show that the filter using the algorithm and objective function proposed in this paper has larger stop band attenuation and smaller ripples compared with the filter using above-mentioned algorithm,which has proved the effectiveness and feasibility of the method.

关 键 词：多组粒子群 分步协同进化 动态调节机制 混沌及高斯动态扰动 云自适应动态调整 FIR数字滤波器 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] TN911.72[自动化与计算机技术—计算机科学与技术]