Job Shop调度风险聚集影响机理与建模方法

There are a lot of uncertainties in job shops, such as machine breakdowns, shortage of raw materials and rush orders, usually they appear at the same time, and they will lead to delay in makespan, increase in cost, reduce in robustness and stability of job shop scheduling. They are the main risks the job shop scheduling has to be faced. The project focuses on the analysis and modeling of risks in job shop scheduling. The methodology of identification and modeling of concurrent risks will be studied by using theory of risk analysis. The identification and clustering methods of the risks in job shop scheduling will be studied, so that the risks can be expressed more accurately and confidently. For the aleatory risk or epistemic risk, the way that the risk affects the job shop scheduling will be analysized, and the evaluation model will be set up. For the risk aggregation, the reason for influence of concurrent risks to the job shop scheduling will be studied, and a combined model for evaluation of concurrent risks will be put forward. A simulation system will be developed based on Teconomatix and MATLAB, and the usefulness of the method will be demonstrated. The project will set up a theoretical foundation for identify and model of risks in job shop scheduling. It will also provide a new method to control the production in job shop.

Job Shop生产中存在的设备故障、原材料供应短缺、紧急到达的工件等不确定因素常常并发，导致生产进度拖延、加工成本上升、调度方案的鲁棒性与稳定性降低，是调度系统面对的主要风险。本项目聚焦于Job Shop调度风险研究领域，对风险聚集的影响机理与建模方法进行探索。分析Job Shop调度风险特征，建立并发风险的辨识方法；分析随机风险和认知风险对调度性能的影响方式，建立风险影响量化评估模型；探讨多风险聚集影响机理，提出随机风险与认知风险对调度影响的叠加方法，建立多风险聚集量化评估模型；建立仿真系统，验证Job Shop调度风险模型与方法的有效性。研究将为Job Shop调度风险分析提供理论基础，为生产调度过程管控提供创新研究方法。

本项目针对Job Shop调度中普遍存在的随机设备故障、加工时间随机变化等随机不确定因素，以及物料短缺、工人熟练程度等认知不确定因素进行了深入系统研究；分别建立了两类不确定因素的风险分析、评价、优化模型，建立了工人熟练程度与加工时间随机变化风险叠加模型，提出了基于调度鲁棒性和稳定性的风险量化评价指标；提出了一种分布估计算法(Estimation of Distribution Algorithm, EDA)和评价空间缩减策略(Reduction Strategy, RS)相结合的优化算法，通过仿真研究与已有算法进行了比较，验证了模型、算法的有效性；基于Teconomatix和MATLAB开发完成Job Shop调度仿真系统，从实际Job Shop中提取仿真数据，在对系统性能进行深入分析的同时，使得仿真结果更具真实性。.项目在《Computers & Operations Research》、《机械工程学报》等国内外刊物及会议发表论文19篇，其中SCI收录10篇、EI收录15篇；项目培养研究生10名，其中1名取得博士学位、5名取得硕士学位，中外合作培养博士研究生2名；项目申报国家发明专利3项，其中1项获授权；项目负责人参加Informs、EURO年会并宣读论文4次。