融合可拓资源的 Job shop 无拖期调度问题研究

It is a core goal for make-to-order manufacturing enterprises to minimize earliness and manufacturing costs with no tardiness. Making full use of manufacturing resources can expedite the realization of the goal. However, it seems that, up to now, no systematical study on no-tardiness job shop scheduling problem (NTJSSP) with consideration of comprehensive utilization of extension resources. To make up the defectiveness, this project mainly focuses on theories, techniques and algorithms on no-tardiness job shop scheduling with consideration of extension resources. Firstly, from the point of view of shop scheduling, we will conduct the analysis and modeling for extension resources by adopting element theory of extenics, Bayesian analysis method, and uncertain linguistic variables comprehensively; Secondly, the time difference analysis technology of AOE net, critical path method, and fuzzy extension evaluation will be used together to generate dynamically extension resources; Further, no-tardiness job shop scheduling problem will be modeled by combining fuzzy extension evaluation with chance-constrained programming model. With respect to the complexity of multidimensional and multistage decisions in NTJSSP, we will integrate multi-mode decisions into a procedure of dynamic optimization iteration to simplify the problem solving. Heuristic algorithm and hybrid non-dominated sorting genetic algorithm will be adopted together to conduct decisions and optimizations for NTJSSP. Finally, a prototype system for solving NTJSSP will be developed with J2EE; simulation scheduling, and system analysis and evaluation for the results will be carried out with the field data collected from die & mould enterprise. The work of this project is significant for both theoretical researches on scheduling and engineering practice.

无拖期前提下最小化提前及成本是面向订单制造企业生产运作关键目标；充分利用制造资源是实现这一目标的重要措施。现有文献尚未见综合利用可拓资源以实现无拖期调度的系统研究。本项目围绕融合可拓资源的Job shop 无拖期调度问题理论与方法展开研究。首先基于调度问题特点，以可拓学基元理论为主导，结合Bayes 分析和不确定性语言变量进行可拓资源分析与建模；其次，采用AOE 网时差分析、关键路径搜索及模糊可拓评价等研究可拓资源动态生成技术；进而将模糊可拓评价与混合机会约束规划相结合构建融合可拓资源的无拖期调度问题模型；针对问题多维多阶段决策特点，采用将多模式决策融入动态优化迭代过程的策略简化问题求解，采用启发式算法与混合非劣分层遗传算法相结合实现问题的决策与优化；最后，以J2EE 技术开发原型系统并基于模具企业现场数据进行模拟调度及结果的系统分析评价。本项目研究对调度理论与方法及工程实际具有重要意义。

制造系统优化调度是实现智能制造的关键技术之一。在优化调度的诸多目标中，保证订单的交货期是最重要和通常考虑的优化目标。由于订单任务具有不确定性，企业资源相对固定，因此，已有关于优化调度的研究在考虑交货期时通常以拖期最小化为目标，而关于保证无拖期的相关研究极少见到。为此，本项目提出融合企业各种可拓资源，通过对相关优化调度技术的研究，以保证订单的无拖期交货。本项目的研究主要集中于以下方面：对考虑作息时间和加班的调度问题、考虑多柔性的调度问题、考虑外协的调度问题和融合各种可拓资源的调度问题进行了分析和建模；提出了求解相应的问题改进的遗传算法与禁忌搜索混合算法、改进自适应遗传算法、基于双层编码的混合遗传算法、改进非支配排序遗传算法、改进遗传算法与模拟退火混合算法、多目标MOEA/D算法、深度强化学习算法、基于规则解码改进遗传算法等算法和相关技术；开发了相应的仿真调度系统，对不同规模和类型的调度问题进行了仿真调度。仿真调度实验表明了所提出技术和算法的有效性。通过本项目的研究，形成了一套较系统的求解无拖期Job Shop调度问题的有效技术方法，不仅对制造系统优化调度理论与方法起到了较好的充实和完善作用，且对企业优化调度系统的开发和实施起到了有效的指导作用。

内容获取失败，请点击重试