CIESC Journal ›› 2012, Vol. 63 ›› Issue (12): 3985-3990.DOI: 10.3969/j.issn.0438-1157.2012.12.035

Previous Articles     Next Articles

Multi-objective optimization based on parallel multi-families genetic algorithm

LU Hai1, YAN Liexiang1, SHI Bin1, LIN Zixiong1, LI Xiaochun2   

  1. 1. College of Chemical Engineering, Wuhan University of Technology, Wuhan 430070, Hubei, China;
    2. Zhenjiang Yctionsoft Co.Ltd., Zhenjiang 212009, Jiangsu, China
  • Received:2012-08-07 Revised:2012-08-17 Online:2012-08-29 Published:2012-12-05
  • Supported by:

    supported by the High-tech Research and Development Program of China(2011AA02A206)and the National High-tech SMEs Technology Innovation Project(11C26213201446).

并行多家族遗传算法解多目标优化问题

卢海1, 鄢烈祥1, 史彬1, 林子雄1, 李骁淳2   

  1. 1. 武汉理工大学化学工程学院, 湖北 武汉 430070;
    2. 镇江雅迅软件有限责任公司, 江苏 镇江 212009
  • 通讯作者: 鄢烈祥
  • 作者简介:卢海(1986-),男,硕士研究生。
  • 基金资助:

    国家高技术研究发展计划项目(2011AA02A206);国家科技型中小型企业技术创新项目(11C26213201446)。

Abstract: A parallel multi-families genetic algorithm (PMOGA) is proposed to reduce computing burden which is incurred in the solution of the multi-objective optimization problem in chemical process when combining the single genetic algorithm (GA) with the process simulator.A master-slave node distributed computing strategy is employed in the proposed algorithm.Based on the idea of decomposition-coordination, the Pareto curve is divided into multi-sections, and then the calculation task of each sub-section is assigned to single computer in LAN to reduce the computing time.The proposed method has been tested on two practical chemical examples.The results show that PMOGA is superior to single GA in both uniformity and comprehensiveness of the Pareto solutions.

Key words: parallel computing, multi-objective genetic algorithm, process simulator, optimization

摘要: 提出了一种并行多家族遗传算法,采用主从节点分布式的计算策略,并应用分解协调的思想,对Pareto前沿进行分段,将计算任务分配到局域网上的多台计算机上完成,以减少计算时间。将所提出的方法用于两个化工实际问题的求解,得到的Pareto前沿的分布均匀性和全面性均优于单个遗传算法算得的结果。解决了遗传算法与流程模拟器结合解化工过程多目标优化问题时计算耗时太长的难题。

关键词: 并行计算, 多目标遗传算法, 流程模拟器, 优化

CLC Number: