基于学习竞争型PSO算法的轧制计划问题求解  

Rolling optimization model based on learning and competitive PSO algorithm

作　　者：黄成[1] 戴蓉[2] 

机构地区：[1]四川工程职业技术学院计算机科学技术系,四川德阳618000 [2]中国民用航空飞行学院计算机学院,四川广汉618307

出　　处：《计算机工程与设计》2016年第7期1836-1840,共5页Computer Engineering and Design

基　　金：四川省教育厅科研基金项目(13ZA0135)

摘　　要：为保证快速准确求解出轧制调度计划的可行解,建立一种含有0-1变量的热轧调度数学模型。针对该模型含有整数变量的特点,提出一种基于学习与竞争的粒子群算法,通过将学习与竞争的思想和PSO算法相结合,让种群中适应值较差的个体以一定的概率向适应值较好的个体进行学习,构成新的粒子群算法迭代表达式,在不增加PSO算法计算复杂度的基础上,克服基本PSO算法的缺点。实验结果表明,该粒子群算法比基本的PSO有着更高的搜索精度,验证了该算法的有效性和准确性。To solve rolling scheduling problem accurately and fast,the mathematical model of hot rolling scheduling problem containing 0-1 variables was established.For solving the optimization model,a particle swarm algorithm based on learning and competition was presented.The learning thought was combined with basic PSO algorithm to make poor populations adapt to the good populations at a certain probability.The new iterative expressions of particle swarm optimization （PSO）can overcome the drawback of basic PSO algorithm without increasing computational complexity.Based on the real data of the factory,experimen-tal results show that the presented particle swarm algorithm has higher search accuracy than the basic PSO.It is effective on solving the hot rolling scheduling problem.

关 键 词：学习与竞争 粒子群算法 热轧调度 0-1变量 适应值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]