基于遗传退火算法的装配线设计多目标优化方法  

GASA-based multi-objective optimization method for assembly line design

作　　者：秦永法[1] 赵明扬[1] 陈书宏[1] 

机构地区：[1]中国科学院沈阳自动化研究所

出　　处：《计算机集成制造系统》2004年第11期1416-1420,1432,共6页Computer Integrated Manufacturing Systems

基　　金：国家973计划资助项目(2002CB3122001)。~~

摘　　要：针对混装配线设计这一有约束的多目标优化问题,建立了数学模型。将基于Pareto的解的分级方法与Lp-范数形式的非线性机制相组合,构建了基于遗传退火算法多目标优化方法。重点阐述了个体编码、染色体检修、多目标处理机制等关键技术。设计了算法流程图,并开发了优化程序。该方法克服了加权和方法的不足,用模拟退火改善了遗传算法全局寻优性能。计算实例表明,随着迭代次数的增加,每代的非受控点逐渐收敛于Pareto最优边界,是一种混装线设计多目标优化的新方法。Assembly line design is a multi-objective optimization problem with constrains. A mathematical model was constructed for the problem of this hybrid assembly line design. A Genetic Algorithm and Simulated Annealing (GASA) based multi-objective optimization method was proposed, in which a Pareto-based solution ranking method and a kind of nonlinear multi-objective tackling mechanism using Lp-norm was included. Some key technologies such as chromosome encoding, the chromosome checking and amending method and the multi-objective tackling mechanism were described in detail. A flowchart of the proposed method was illustrated, and an optimization program was developed. The method overcame the deficiency of the weighted sum method and improved the searching performance of GA. Computing example shows that the non-dominated points of evolutionary generations are convergent eventually to the Pareto optimal frontier. It provides a new method for optimizing more than two design objectives such as workload balance, resource planning and the system reliability at one time.

关 键 词：装配线设计 多目标优化 Pareto最优化 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]