基于参考点和差分变异策略的高维多目标冷轧负荷分配  

Many-objective Load Distribution for Cold Rolling Based on Reference Point and Differential Mutation Strategy

作　　者：赵志伟[1] 侯宇浩 王伟志[1] 么洪波 

机构地区：[1]唐山学院计算机科学与技术系,河北唐山063000 [2]燕山大学电气工程学院自动化系,河北秦皇岛066004 [3]河北中瓷电子科技有限公司,河北石家庄050051

出　　处：《计量学报》2017年第6期730-734,共5页Acta Metrologica Sinica

基　　金：河北省科技计划项目(16212302);河北省高等学校科学研究青年基金项目(QN2017416);唐山学院博士创新基金

摘　　要：为提高高维多目标进化算法收敛精度,分析了参考点及2个差分变异策略的特性,提出了RN2-DE算法改进变异策略,用以替代高维进化算法R-NSGA-II的变异策略。4维函数DTLZ2的实验结果表明,所提算法收敛精度得到明显改善。通过等功率裕量、最小轧制功率、预防打滑、末机架板形良好4个目标函数建立冷轧负荷分配的高维多目标优化问题,并使用所提算进行优化。To improve the accuracy of many-objective evolutionary algorithm, the characteristics of the reference point and two differential mutation strategies are analyzed. The RN2-DE algorithm is proposed to improve the mutation strategy to replace the mutation strategy of R-NSGA-II. The experimental results of function DTLZ2 for 4 dimensions shows that the proposed algorithm significantly improves the convergence accuracy. Many-objective optimization problem of load distribution for cold rolling is established by four objectives, such as equal power margin, minimum rolling power, slipping preventing and good flatness of the final stand, and is optimized by the proposed algorithm.

关 键 词：计量学 负荷分配 冷轧 高维多目标 变异策略 

分 类 号：TB93[一般工业技术—计量学]