文献类型：期刊文献

中文题名：基于高斯变异的智能单粒子算法

英文题名：Improved ISPO algorithm based on Gaussian mutation

作者：夏伟[1];程慕鑫[1];刘漫丹[1]

机构：[1]华东理工大学化工过程先进控制和优化技术教育部重点实验室,上海200237

年份：2013

卷号：30

期号：4

起止页码：986

中文期刊名：计算机应用研究

外文期刊名：Application Research of Computers

收录：CSTPCD;;北大核心:【北大核心2011】；CSCD:【CSCD2013_2014】；

基金：广东省产学研资助项目(2010B090400477)

语种：中文

中文关键词：粒子群优化;智能单粒子算法;高斯变异

外文关键词：particle swarm optimization; intelligent single particle optimizer（ISPO）; Gaussian mutation

摘要：针对智能单粒子优化算法(ISPO)容易出现算法早熟、收敛精度低的现象,提出一种基于高斯变异的智能单粒子算法(GISPO)。当粒子陷入局部最优值,每一维速度会降到一定的阈值,整个粒子进化处于缓慢阶段;此时给予搜索到的历史最优极值一个自适应的高斯变异扰动,会大大提高粒子的逃逸能力,帮助粒子快速地跳出局部极值点,不断地向全局最优解靠近。通过几个标准测试函数进行实验,结果表明该算法的收敛速度、搜索精度和稳定性均优于ISPO算法。
The intelligent single particle optimizer（ISPO） has some shortcomings such as prematurity and low convergence accuracy.This paper proposed an improved ISPO algorithm based on Gaussian mutation to overcome the problem.When the particle fell into local optimal point,each speed of the particle dropped to a certain threshold,and the particle evolved very slowly.When an adaptive Gaussian mutation disturbance was added to the history optimization solution,it would improve the particle＇s escape ability greatly and help particle to jump out of local optimal point.In this way,the particle can get close to global optimal point quickly.The optimal results on benchmark functions show that the improved ISPO algorithm based on Gaussian mutation can converge more quickly and get better stability.