基于定界多目标优化的区别化生产调度问题研究

This project proposes a new type of scheduling called the problem with the distinctness of orders and focuses on the researches on the problem and its delimited multiobjective optimization because of the risks and situations leading to the frequent delayed delivery in the production process of make-to-order company. Firstly, the delimited multiobjective optimization theory is considered. The mathematical model of the scheduling problem with the distinctness of orders is obtained after all orders are classified by considering factors such as specification and revenue and objectives are categorized into the primary ones and the secondary ones and the latter are delimited according to their relative importance. Secondly, to effectively solve several problems with the distinctness of orders, in which resources state of company or the features of orders are considered, the features of some constraints and conditions such as resources selection, outsourcing, material and financial constraints, blocking, assembly and reentrant are utilized, a new algorithm called controlled local search is proposed, and the delimited multiobjective optimization methods based on controlled local search and imperialist competitive algorithm are systematically studied. Finally, for the scheduling problems with the distinctness of orders and disruptions, the disruption-caused objective is defined and delimited after the negative effects of disruptions on the original schedule are measured, and the reasonable and effective disruption management strategies are studied. This project will provide the theory and methods on scheduling with the distinctness of orders and extend the research contents of production scheduling, disruption management and multiobjective optimization. This project also can provide the decision supports for the high-level performances of make-to-order manufacturing company on on-time delivery and resource utilization et al.

本项目针对MTO制造企业生产过程中存在的各种导致延期交货经常发生的状况与风险，提出新型调度形式—区别化调度问题，开展关于该类问题及其定界多目标优化的研究。首先，研究定界多目标优化理论，根据订单规格和收益等对订单分类,根据目标相对重要性确定主要目标和次要目标并对后者定界，建立问题的数学模型；其次，针对多种考虑企业资源状况或订单特点的区别化调度问题，结合资源选择、生产外包、原材料与财政约束、阻塞、装配和可重入等约束与条件的特点，提出新算法—受控局部搜索，并系统研究基于该算法或帝国竞争算法的定界多目标优化方法；最后，针对面向干扰事件的区别化调度问题，在度量干扰事件对原调度的负面影响的基础上，定义并定界扰动目标，研究合理高效的干扰管理新策略。本项目研究将提供一套区别化调度理论与方法，扩展生产调度、干扰管理和多目标优化的研究内容，为MTO企业获得按时交货和资源利用等方面的高水平性能提供决策支持。

项目围绕MTO制造企业生产过程中经常存在的各种导致延期交货发生的状况和风险，在定界多目标优化理论基础上，开展区别化生产调度问题研究，根据订单规格型号和收益等对订单分类，将目标分为主要目标和次要目标并对次要目标定界，针对考虑企业资源状况、订单特点和干扰事件的区别化调度问题，应用受控局部搜索和帝国竞争算法等对问题进行定界多目标优化。目前已在国际TOP和重要SCI期刊IEEE TCYB, CAIE和EAAI等发表论文16篇，在自动化学报等国内重要期刊发表论文18篇。这些成果SCI引用次数超过150次，入选ESI高被引论文一篇。项目研究期间负责人连续四年入选Elsevier中国高被引学者榜。在该项目资助下，培养博士生2名，硕士生10名。.项目在定界多目标优化与区别化调度方面，以总能耗这一具体目标为例，研究了总能耗不超过给定阈值、总能耗重要性低于其它目标、总能耗为次要目标而总延迟时间等为关键等情形，其中IEEE TCYB论文将总能耗定为次要目标并对其定界，给出了一种有效的双阶段过程，第1阶段同等对待所有目标，而第2阶段直接优化问题本身，并应用帝国竞争算法对问题求解。在区别化多资源调度方面，将这类问题从单工厂扩展到多工厂，并加入能源等资源，深入研究了绿色柔性作业车间调度、绿色混合流水车间调度、分布式并行机调度和分布式混合流水车间调度，目前围绕区别化多资源调度在EAAI、CAIE、IJPR和自动化学报等期刊发表17篇论文，取得了一些全新结果。在区别化装配调度方面，研究了II型装配线平衡问题和装配混合流水车间调度问题。除了这些区别对待优化目标的研究外，本项目还研究了区别对待工件的区别化调度问题。发表在IJPR上的论文针对考虑订单外包的作业车间调度问题，这是一类MTO制造环境下常见的问题，应用一种性能优良的蛙跳算法优化总延迟时间和总外包成本，双主体区别化调度也得到研究。此外，帝国竞争算法是项目重点研究的方法，发表了帝国竞争算法在区别化调度等方面的应用论文9篇，验证了该算法的搜索优势，推动了该算法的研究进展。