工业工程与管理
工業工程與管理
공업공정여관리
Industrial Engineering and Management
2015年
5期
143-150
,共8页
许绍云%王雷%李铁克%刘玉琢
許紹雲%王雷%李鐵剋%劉玉琢
허소운%왕뢰%리철극%류옥탁
热轧批量调度%多目标优化%基于种群修复的改进强度 Pareto 进化算法%圆钢生产
熱軋批量調度%多目標優化%基于種群脩複的改進彊度 Pareto 進化算法%圓鋼生產
열알비량조도%다목표우화%기우충군수복적개진강도 Pareto 진화산법%원강생산
hot-rolling batch scheduling%multi-objective optimization%population repair based improved strength pareto evolutionary algorithm%round steel production
针对圆钢热轧批量调度问题,考虑热轧工艺约束、综合生产需求优化等因素,建立了以最小化最大完工时间、机器调整成本以及提前\拖期时间为目标的优化模型,并构造了基于种群修复的改进强度 Pareto 进化算法。算法以改进强度 Pareto 进化算法为框架实现批量序列调整,并结合问题变量的离散性特征设计编码规则,以及选择、交叉和变异等遗传进化方式;考虑问题约束的特殊性,设计种群修复策略,以修正违反热轧工艺约束的批量序列。通过基于实际生产数据的仿真实验验证了模型和算法的可行性和有效性。
針對圓鋼熱軋批量調度問題,攷慮熱軋工藝約束、綜閤生產需求優化等因素,建立瞭以最小化最大完工時間、機器調整成本以及提前\拖期時間為目標的優化模型,併構造瞭基于種群脩複的改進彊度 Pareto 進化算法。算法以改進彊度 Pareto 進化算法為框架實現批量序列調整,併結閤問題變量的離散性特徵設計編碼規則,以及選擇、交扠和變異等遺傳進化方式;攷慮問題約束的特殊性,設計種群脩複策略,以脩正違反熱軋工藝約束的批量序列。通過基于實際生產數據的倣真實驗驗證瞭模型和算法的可行性和有效性。
침대원강열알비량조도문제,고필열알공예약속、종합생산수구우화등인소,건립료이최소화최대완공시간、궤기조정성본이급제전\타기시간위목표적우화모형,병구조료기우충군수복적개진강도 Pareto 진화산법。산법이개진강도 Pareto 진화산법위광가실현비량서렬조정,병결합문제변량적리산성특정설계편마규칙,이급선택、교차화변이등유전진화방식;고필문제약속적특수성,설계충군수복책략,이수정위반열알공예약속적비량서렬。통과기우실제생산수거적방진실험험증료모형화산법적가행성화유효성。
In this paper,a hot-rolling batch scheduling problem of round steel is studied. Considering the production technology constraints and production integration objectives,a multi-objective integer programming model is built with the objectives to minimize the makespan,the setup costs and earliness\lateness of orders,and a Population Repair based Improved Strength Pareto Evolutionary Algorithm is developed to solve the model.In the proposed algorithm,the basic Improved Strength Pareto Evolutionary Algorithm is applied to adjusting the batch sequences,and then encoding rule and genetic evolution methods,such as selection,cross-over and mutation,are designed corresponding to the discreteness of variables in the model.Considering the specificity of problem constraints,some population repair strategies are also proposed to rearrange batch sequences those contrary to hot-rolling process.Experiments with practical production data show that the model and algorithm are feasible and effective.