3D打印制造模式下订单分批与动态调度优化方法研究

3D printing is gradually providing a new manufacturing mode for the traditional manufacturing industry to meet the needs of personalized and complex manufacturing. This new manufacturing mode has brought great changes to the production methods of modern enterprises, especially in the field of production scheduling. 3D printing manufacturing process has several difficulties such as non-serial (parallel) batch processing type, three-dimensional constraints, unpredictability, etc. As a result, a new challenge is posed for the classical production scheduling theory and methods. In this project, the main contents contain the following aspects. Based on data analytics, we estimate key parameters affecting the batch scheduling performance. Under the three-dimensional constraints, we construct models of the batch scheduling based on parameter estimation, robust optimization based on uncertain parameter information, integrated optimization of real-time order acceptance and dynamic batch scheduling. After analyzing the model properties, we respectively design the fast algorithms based on branch and bound framework, exact iterative algorithms based on decomposition, bilevel optimization algorithm based on column generation and neighborhood search. Finally, we carry out the applied research. This project is anticipated to make innovative contributions in theories and methods of batch scheduling in 3D printing manufacturing mode on one hand, and help the 3D printing enterprises to promote production management level, make the shift of manufacturing industry, and meet the strategic requirement of green and intelligent manufacturing on the other hand.

3D打印正逐渐为传统制造应对个性、复杂制造需求提供一种新型加工模式，这一加工制造模式给现代企业的生产方式带来重大变革，特别是生产调度领域首当其冲。3D打印生产过程兼具批加工类型的非串（并）行性、三维空间约束性、产品不可预知性等难点，对经典生产调度理论与方法提出了新挑战。主要研究内容包括：分析影响批调度性能的关键参数并进行科学估计，在打印机三维空间容量限制下，构建基于参数估计的分批与调度、基于不确定参量信息的鲁棒优化、实时订单接受和动态批调度的集成优化模型；通过模型转化与性质分析，分别设计基于分支定界框架的快速求解算法、基于分离思想的精确型迭代算法以及基于列生成和大邻域搜索的双层优化算法；最后进行数值检验并开展应用研究。以期在3D打印制造模式下的订单分批与调度优化理论与方法上取得创新性成果，提升3D打印制造企业生产管理水平，促进制造业转型升级，符合绿色制造、智能制造的战略需求。

作为制造业的一个新兴领域，增材制造（又称3D打印）的发展和普及给社会生产和生活带来重大变革。然而，增材制造企业一直面临工时长、制造成本高、装备成型尺寸受限等问题。如何根据增材制造模式的生产特征，克服其内在的复杂性，提供科学有效的生产调度优化方法，已成为产业界和学术界共同关注与研究的热点和难点。按照项目申请书中拟定的研究方案，目前项目组完成了如下工作：（1）针对增材制造与传统制造工艺相结合的产品可制造性优化问题，以总生产时间、总成本和服务质量为目标建立了多级别制造商协同生产任务分配模型，设计了基于切比雪夫分解策略的MOEA/D算法。（2）针对订单取消风险下的生产资源预留与安排问题，构建了相应的随机规划模型，提出基于列生成分解策略的启发式算法。（3）针对实时订单动态到达，研究了订单接受和动态调度的集成优化问题，将其转化为马尔科夫决策过程，设计了集成接单和派单两个智能体的深度强化学习算法。（4）增材制造生产流程具有流水车间模型的特征，流水车间的子模型为单机调度模型，因此深入研究了具有恶化效应、交货期窗口或外包等特征的基础单机调度模型。综合而言，整个研究进行了一些创新性的探索，将优化调度方法延伸至增材制造生产系统，为实施增材制造的企业制定生产计划与调度方案提供理论、方法和技术支持。