数学优化和人工智能助力智能制造生产线--基于上汽大众新能源汽车生产的案例研究  

Combining Mathematical Optimization and Artificial Intelligence to Improve Intelligent Production Lines--A Case Study on SAIC Volkswagen’s New Energy Vehicle Production

作　　者：吴泽锐 刘冉[1] 陈晓东 易延洪 WU Zerui;LIU Ran;CHEN Xiaodong;YI Yanhong(School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China;SAIC Volkswagen,Shanghai 201805,China)

机构地区：[1]上海交通大学机械与动力工程学院,上海200240 [2]上汽大众汽车有限公司,上海201805

出　　处：《工业工程与管理》2021年第6期208-218,共11页Industrial Engineering and Management

摘　　要：汽车行业的快速发展,市场个性化需求的凸显,对汽车企业的多车型生产能力提出了挑战。上汽大众的新能源汽车MEB工厂具备多车型混流生产能力,但受限于产线上的物料空间与生产能力,不能随时任意切换生产车型,需要结合批量制造以降低生产成本。通过对该工厂含"最小批量"的多车型生产决策案例进行研究,发现数学优化与人工智能相结合的决策优化技术能够有效助力智能制造。研究结论表明,面向排产计划场景可以建立混合整数规划模型或设计启发式优化算法,而运用强化学习方法则可以有效应对实时调度决策问题。研究结论对当前我国制造业智能化转型升级具有重要启示。Automotive firms face significant challenges in producing multi-models due to the rapid development of the automotive industry and the emergence of individualized market demands.SAIC Volkswagen′s MEB plant for new energy vehicles is capable of multi-model mixed-flow production.Because of the limited part space and production capacity,the company could not change production models at any time and must combine mass production to reduce production costs.In the study of the factory′s decision-making case for multi-model production with“minimum batch size”,we found that combining mathematical optimization and artificial intelligence could help with intelligent manufacturing.The research concludes that a mixed-integer programming model or heuristic optimization algorithm can perform well for the production scheduling scenario,and that reinforcement learning methods can effectively deal with real-time decision-making problems.The research conclusions have profound implications for the current intelligent transformation and upgrading of China′s manufacturing industry.

关 键 词：智能制造 数学优化 人工智能 汽车制造业 

分 类 号：F224[经济管理—国民经济]