一种求解矩形排样问题的遗传-离散粒子群优化算法  

A Genetic-Discrete Paticle Swarm Optimization Algorithm for Rectangular Packing

作　　者：黄岚[1] 齐季 谭颖[1] 杨滨[1] 

机构地区：[1]吉林大学计算机科学与技术学院,吉林长春130012 [2]总参陆航研究所,北京101121

出　　处：《电子学报》2012年第6期1103-1107,共5页Acta Electronica Sinica

基　　金：国家自然科学基金(No.60873146;No.60973092;No.60903097);"符号计算与知识工程"教育部重点实验室项目;吉林省科技发展计划重点项目(No.20110354)

摘　　要：针对制造业领域的矩形优化排样问题,提出一种遗传-离散粒子群优化算法.引入交换子和交换序概念,解决了标准粒子群优化算法在求解组合优化问题时粒子的更新难以描述问题;融合遗传算法的交叉与变异思想,增强了粒子群的多样性和稳定性;同时采用改进的最低水平线搜索算法加快算法的收敛速度,并解码形成排样方案.通过实验数据对比,验证了该算法在求解矩形排样问题中的高效性和鲁棒性.For the optimization problem of rectangular packing in manufacturing field, a geneticdiscrete particle swarm opti mization is proposed. Through introducing the concepts of exchange operator and exchange sequence, the difficult problem of de scribing the update particles in the standard particle swarm optimization algorithm for combinatorial optimization problem is solved. Combining the idea of mutation and crossover in genetic algorithm strengthens the diversity and stability of the particle swarm. An improved lowest horizontal search algorithm is proposed to accelerate the convergence speed and can be used to decode to packing layout. The experimental data show that the proposed algorithm is effective and robust in solving the problem of rectangular pack ing.

关 键 词：离散粒子群优化 遗传算法 最低水平线搜索 矩形排样 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]