基于多智能体仿真建模的板坯到库预测  

Slab stockyard arrival prediction based on multi-agent simulation modeling

作　　者：文静 林海[2] 马乾升 WEN Jing;LIN Hai;MA Qiansheng(Research Institute,Baoshan Iron&Steel Co.,Ltd.,Shanghai 201999,China;Products&Technique Management Department,Baoshan Iron&Steel Co.,Ltd.,Shanghai 201999,China)

机构地区：[1]宝山钢铁股份有限公司中央研究院,上海201999 [2]宝山钢铁股份有限公司制造管理部,上海201999

出　　处：《宝钢技术》2025年第1期40-44,共5页Baosteel Technology

摘　　要：针对热轧板坯资源不平衡的问题,需要结合炼钢计划对板坯到库时间进行动态预测,从而有预判性地承接来料,为热轧计划的制定和整体物流平衡提供有力支撑。基于多智能体技术,建立包含连铸机智能体、物流管控智能体和数据交互功能模块的板坯到库仿真模型,能够对设备加工及板坯物流等动态过程实现较高精度的模拟和预测。采用历史实绩数据驱动仿真模型运行,试验结果表明板坯切断时间预测偏差在10 min之内的比例为84.59%,辊道直送的板坯到库预测偏差在30 min之内的比例为89.69%。To address the issue of uneven distribution of hot-rolled slab resources,dynamic prediction of slab-to-stockyard time based on the steelmaking schedule is necessary.This proactive approach enables anticipatory material handling,thereby providing robust support for hot rolling schedule formulation and overall logistics balance.Using multi-agent simulation technology,a simulation model was developed incorporating intelligent agents for continuous casting machines,logistics control,and data exchange modules.This model achieves highly accurate simulation and prediction of dynamic processes such as equipment processing and slab logistics.Driven by real production data,experimental results demonstrate that slab cutting time prediction deviates within 10 min in 84.59%of cases,while slab-to-stockyard prediction for direct conveyance deviates within 30 min in 89.69%of instances.

关 键 词：板坯物流 到库预测 多智能体 仿真模型 

分 类 号：TG338[金属学及工艺—金属压力加工] N945.13[自然科学总论—系统科学]