多目标中间包计划模型及混合优化算法  

Multi-objective tundish planning model and hybrid optimization algorithm

作　　者：马天牧[1] 罗小川[1] 柴天佑[1,2] 

机构地区：[1]东北大学流程工业综合自动化国家重点实验室,辽宁沈阳110819 [2]东北大学自动化研究中心,辽宁沈阳110819

出　　处：《系统工程学报》2015年第4期451-465,共15页Journal of Systems Engineering

基　　金：国家自然科学基金资助项目(71021061;60974091;61174187;61104174);国家973重点基金研究发展计划资助项目(2009CB320601);高校学科创新引智计划资助项目(B08015)

摘　　要：中间包计划是重要的炼钢-连铸批量计划之一,其功能是在给定的炉次计划中,根据炼钢-连铸生产能力及下游工序生产所需材料的数量,挑选出待生产的炉次并制定出合理的中间包使用数量及每个中间包内生产的炉次.在对中间包计划问题描述的基础上及现有文献中未考虑中间包利用率及中间包内炉次宽度差异性,建立了多目标中间包计划数学模型.为了求解模型将模型分解为两个子模型,并针对两个子模型设计了迭代局部搜索算法(iterated local search,ILS)及变邻域搜索算法(variable neighborhood search,VNS)相结合的双层混合算法,考虑到中间包利用率及多目标权重对解的影响,在算法中加入了可调整模型参数的方法,最后用实际生产数据对模型及算法进行验证.Tundish planning is one of the most fundamental batch planning problems in steelmakiag continuous casting. It determines how many tundishes are used to produce charges regarding the capacity of steelmaking and the quantity required by the downstream. Based on the description of the tundish planning problem and the shortcoming of existing tundish planning methods in the literature, a multi-objective mathematical model is built. The model was divided into two sub-models for solving, and a two-layer hybrid algorithm is presented by combining the iterated local search with variable neighborhood search. Considering the utilization of tun（fishes and the weights of multiple objectives, two methods for adjusting the parameters of the algorithm were introduced. Finally, real-world data are used to test the model and algorithm. The results show the effectiveness of the model and algorithm.

关 键 词：中间包计划 多目标 迭代局部搜索 变邻域搜索 混合算法 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]