多目标柔性作业车间调度的多交叉策略元胞进化算法  

Multi-crossover Strategy of Multi-objective Cellular Genetic Algorithm Research on Flexible Job-shop Scheduling Problem

作　　者：林震[1] 帅剑平[1] 袁煜[1] 

机构地区：[1]桂林电子科技大学教学实践部,桂林541004

出　　处：《科学技术与工程》2017年第7期69-76,共8页Science Technology and Engineering

基　　金：国家自然科学基金(61261017);广西无线宽带通信与信号处理重点实验室项目(GXKL061501);广西教育厅广西高校科学技术研究项目(KY2015ZL040)资助

摘　　要：针对经典Job-shop调度问题的局限性,构建了以加工成本、瓶颈机器负荷、机器总负荷及制造工期为目标函数的柔性作业车间调度多目标优化模型,提出了基于多交叉策略的元胞多目标遗传算法。在分析优化模型的基础上,使用双层编码方式,并采用多个交叉算子协同进化,提出一种多交叉策略的进化算子。针对元胞多目标遗传算法的特点,提出一种改进的精英策略,保证更多的精英个体参与进化,从而提升算法收敛速度。通过2个基准实例求解对比分析,表明所提方法的有效性。将新算法应用于实际生产企业的车间调度问题中,得到了一组Pareto解集,并采用层次分析法得到一种满意度最大的方案。数据结果表明,该算法在解决多目标FJSP的工程有效性。Aiming at the limitations of classical job-shop scheduling problem,a multi-objective flexible job-shop scheduling optimization model is put forward which production cost,workload of the bottleneck machine,total workload of machines and the make span commonly concerned in complicated manufacturing system are considered.A multi-crossover strategy of multi-objective cellular genetic algorithm is proposed for MOFJSP.According to the complexity of the constructed model,an improved cellular genetic algorithm with multi-crossover strategy is proposed.Based on the analysis of the model,the binary level coding scheme is adopted and several crossover operators can work in coordination.According to the characteristics of the cellular genetic algorithm,an improved elitist strategy is proposed to ensure more elitist individuals in evolution to speed up the convergence of the new.The effectiveness of the proposed algorithm is validated through comparative analysis of two instances.Therefore,the new algorithm is adopted for a practical job-shop scheduling problem and a Pareto set is obtained.The analytic hierarchy process is utilized for find out the maximum satisfactory solution from the obtained Pareto set.The computation results prove its effectiveness of the new algorithm for solving multi-objective flexible job-shop scheduling problem.

关 键 词：柔性车间调度 元胞结构 多交叉策略 层次分析法 多目标优化 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]