【作者】 卢克斌；

【导师】 汪定伟；

【作者基本信息】 东北大学 ， 系统工程， 2010， 博士

【摘要】 随着全球化市场竞争的日益加剧,顾客的需求日益多品种小批量化而且对交货期的要求越来越严格,钢铁企业正面临着越来越严峻的挑战,因此,钢铁生产中的生产计划、性能预报、质量管理等问题引起了学术界的广泛关注。炼钢-连铸在钢铁生产的整个工序中的地位十分关键,很多文献对炼钢-连铸的生产计划问题展开了研究,然而由于钢铁生产的复杂性,很多研究结果在生产实践中只能作为参考。针对炼钢-连铸中的生产计划问题尽管已经提出了很多研究方法,但是应用基于仿真的优化的研究还未见报道。因此,本文应用基于仿真的优化等方法对炼钢-连铸生产计划中的若干关键问题展开了研究,主要包括如下一些内容：(1)查阅了大量的专著、期刊、会议论文集、研究报告等相关文献,对相关领域的研究现状进行了综述,包括：钢铁生产的工艺流程与主要管理模式；钢铁生产管理中的物料采购与优化、生产计划与调度、生产质量管理等问题,并着重综述了生产计划与调度方面的研究；还综述了主要的相关研究方法,并着重介绍了继数学规划和智能优化方法之后出现的基于仿真的优化(Simulation Based Optimization,简称SBO)的发展现状、主要理论工作及应用领域等。(2)提出了炼钢组炉问题的一个多目标优化模型并开发了求解算法。该模型针对单个用户的需求量可能大于单炉容量的实际需求,以余材量和完成合同需要的总成本为目标,其中考虑了钢级之间的替代成本。对模型进行分析和化简之后,开发了一个嵌入禁忌表的求解算法。最终的仿真算例表明了模型和算法的有效性,并分析了余材量和合同需求量之间的关系,这种方法得到的非劣解分散性较好,能够为决策者提供较大的决策空间。(3)借助图论的方法建立了组浇计划的一个数学模型并开发了基于遗传算法的求解算法。假设所有炉次都必须安排浇次、浇次数目未知、所有连铸机相同,该组浇问题被描述成为一个多城市类型且城市间距离不对称的多旅行商问题,目标函数极小化旅行商的总行程和参与的旅行商数目,单个旅行商访问的城市有最大数目限制。开发了一个带分隔符的顺序编码、单切点顺序交叉的遗传算法,并通过随机生成的算例对算法的性能进行了测试,讨论了与成本相关的参数的设定方法,分析了参数对结果的影响,以及将SBO方法引入其中的必要性与可行性。(4)开发了作业计划的一个基于仿真的遗传算法。借助数学模型对炼钢-连铸作业计划(调度)问题进行了描述,其中以连铸机的最早可开浇时刻与实际开浇时刻的最大间隔最小化为目标。基于SBO方法开发了一个求解算法,其中优化器采用分段顺序编码的遗传算法,简洁明了易于遗传操作,仿真器不仅仿真实际系统的运行,也仿真现场人员的一些操作,仿真速度较快。源于文献的仿真案例表明这种方法便于考虑实践中的不确定因素,能使得作业计划更贴近生产实践。(5)设计、开发了一个炼钢-连铸集成调度系统并应用于生产实践。该系统采用客户端/中间件/服务器的三层体系结构,其中中间件基于TUXEDO开发,核心功能模块包括调度的确定与调整、计划的匹配与优化、异常物流的处理等,核心算法采用“人机交互+优化算法”的模式。系统具有如下特点：可与MES系统无缝集成、人机界面友好、可维护性好。此外,本文还提出了一个基于SBO方法研究炼钢-连铸中长期计划的框架,包括计划期精细粒度的划分问题、不同计划之间的接口问题、计划的评价体系等。更多还原

【Abstract】 The iron and steel enterprises are facing stronger and stronger challenges as the global completion in market is enhanced. Meanwhile, the customers demand small lot with different varieties and strict delivery time windows. Therefore, more and more attentions are paid to the issues such as production planning, productivity forecast and quality management in the iron and steel corporations. The steelmaking and continuous-casting (SCC) production is very important in all the processing stages of steel production. As a result, many types of production planning problems in the SCC production are researched in the literature. However, the majority of the results in the literature cannot be applied and only be referenced in actural application because the SCC production process is very complex. In spite of variety approaches that are developed for the production planning problems in the SCC process, the application of the simulation based optimization (SBO) has not been found until now.Therefore, several key problems in the SCC process planning are researched using the approaches including the SBO approach in this thesis. The main researches are as follows:(1) A large quantity of monographs, journal articles, proceeding papers and reports are carefully checked. The following related fields are reviewed:the whole sequence of processing stages in the iron and steel productin; the existing management modes; the management problems in the iron and steel production such as materials purchase and optimization, production quality management and particularly production planning and scheduling. Also, the related approaches, especially the development, theoretical and applied researches of the SBO approach are summarized.(2) A multi-objective mathematical model of the charge batching problem in steclmaking is proposed with its solution algorithm developed. The model minimizes the open ordered slabs and the total costs including the substitute costs between grades respectively with the assumption that the weight of a single customer is perhaps greater than a charge. A heuristic algorithm with an embedded tabu list is developed after the analyses and simplification of the model. The final experiments indicate the validity of the algorithm. Also, the relationship between the obtained open ordered slabs and the given customer weights is analyzed. The Pareto solutions scatter well, which provides great space for the decision makers.(3) A graph-based mathematical model for the cast batching problem is built and solved based on the genetic algorithm (GA). It is assumed that all the charges must be assigned, that number of casts is unknown, and that all the casters (tundishes) are homogeneous. The problem is formulated as a multi-city-type asymmetric multiple traveling salesmen problem, which minimizes both the total traveling distances and the number of involved traveling salesmen with the constraint of maximum number of cities that can be visited by one salesman. A GA based algorithm is developed. The algorithm uses an ordered encodeing with separators and ordered crossover with single corss point. The GA algorithm is tested based on a number of randomly generated instances with the valuing method of parameters discussed. Finally, the effects of paremeters on the results and the necessity and feasibility of introducing the SBO method are analyzed.(4) A simulation based GA algorithm for the job scheduling problem is developed. With the help of mathematical model, the job scheduling problem in SCC process is formulated, the objective function of which minimizes the maximum gap between the earliest available time and starting casting time of casters. The problem is solved based on the SBO method. The optimizer of the SBO is a segmented ordered encoding GA, which is easy for genetic operation. The simulator of the SBO simulates not only the actual system but also the operators, which runs fast. Finally the experiments show that it is convenient to consider the uncertain factors in this method and that the obtained scheduling is much closer to the actual system.(5) An integrated scheduling system for the SCC process is designed, developed and applied. The system is based on the three-layer architecture as client, middle ware, and server. The middle ware is realized using TUXEDO. The main modules of the system consist of scheduling and re-scheduling, match and optimization of planning, and the handling of abnormal logistics. The core algorithm is the hybrid of optimization and human-interactive method. The system has such features as friendly human-machine interface, seamlessly integration with the MES system, and easy maintenance. Furthermore, the author proposes a research frame work of SBO for the medium- or long-term planning in the SCC production, including the division of planning period, the interface between plans and the evaluation system.更多还原