【作者】 何正为；

【导师】 唐任仲；

【作者基本信息】 浙江大学 ， 机械制造及其自动化， 2013， 博士

【摘要】 信息技术的发展推动着企业生产方式的变革。越来越多的中小离散制造企业希望应用信息化手段来管理车间生产现场,实现提高企业生产效率、降低生产成本和提高产品质量的目标。然而企业信息化深入到车间生产管理时,经常会遇到很多问题。论文以构建车间生产现场驾驶舱控制系统为目标,使得管理者可以不必亲临生产现场,通过驾驶舱就能获取车间现场的实时状态,并可以控制生产过程的运行。构建驾驶舱控制系统需要解决三个关键技术：车间现场生产状态的建模、车间生产状态的偏差分析以及车间动态生产调度。这三个关键技术也正是车间生产管理信息化所面临的问题。车间现场生产状态的模型是研究其他关键技术的基础。车间生产现场的基本构成要素是工件和工作单元,两者通过工序联系在一起。三者的结合体抽象成信息节点,分析了信息节点之间的关系,并对信息节点的属性和行为作了描述。以信息节点为基本单元,描述了车间现场的生产状态。分析了车间现场的生产数据与工件状态、工序状态和工作单元的状态的关系RSD,以及信息节点状态与工件状态、工序状态和工作单元的状态的关系RSI,建立了车间现场生产数据与信息节点状态的对应关系,建立了车间现场生产状态的模型。应用信息节点的模型,简化了车间现场生产状态的建模。车间现场理想的生产状态由基准生产过程确定,基准生产过程由作业计划调度确定。实际的生产过程按照基准生产过程确定的节点激活顺序运行,并期望实际生产过程与基准生产过程保持一致。但由于车间生产现场存在大量的随机因素,使得实际生产过程经常偏离基准生产过程,形成实际生产状态与理想生产状态的偏差e(t)。只有同质才能比较,因此把偏差e(t)细化成生产时间偏差和生产成本偏差等,并给出了计算方法。通过引入信息节点关联模块的概念,分析了偏差e(t)在车间生产过程中传播的机理,这是分析预估偏差的基础。实际的车间生产过程是动态变化的,因此需要不断的重调度才能满足实际生产的需求。动态生产调度以提高企业生产效率、降低生产成本和提高产品质量为目标,是多目标优化问题。为了把多目标优化问题转换成单目标优化问题,提出了当量成本的概念。通过分析时间、成本和质量的关系,把时间、成本和质量转换成当量成本,并以当量成本最小作为调度问题的目标函数,把生产状态的偏差不超过设定的阈值作为约束条件。根据车间现场的生产状态,分析了几种触发动态生产调度的典型情况,针对这些情况的特点,给出合适的调度策略和调度算法,实现当量成本最小的调度目标。动态生产调度的实时性要求调度算法必须高效,论文结合启发式规则和邻域搜索的高效性,以时间邻域内的最优解作为启发式搜索方向,提出了信息节点分段寻优算法。在此研究的基础上,对驾驶舱控制系统的体系结构进行了研究,并结合实证案例对论文的理论和方法进行了阐述。更多还原

【Abstract】 Development of Information Technology has been promoting the changes of Production Mode. More and more small and medium discrete manufacturing enterprises want to achieve the goals of improving the production efficiency, reducing the production costs and improving the products’quality by means of the workshop-production-field management with Information Technology. However, many enterprises encountered lots of problems as the Information Technology introduced into the workshop-production-field management thoroughly. To build a remote lean cockpit is the goal of this paper. The controller, without having to visit the workshop-production-field, can get the workshop production state and control the operation of the production process remotely with the lean cockpit.There are three key technologies, which include the modeling of the workshop production state, the analysis of the deviation of the production state and the dynamic production scheduling of the workshop-production-field, to be solved to build the remote lean cockpit. The three key technologies are the problems faced by the production management with Information Technology.The model of the workshop production state is the basis to research on the other key technologies. The fundamental elements of the workshop-production-field include the workpiece and the workstation. The relation between the workpiece and the workstation was built through the working procedure. Information node was the abstraction of the combination of the workpiece, the working procedure and the workstation. The relation between the information nodes has been represented, and the attributes and operations of the information node have been set forth. The workshop production state has defined on the basis of information node as a fundamental unit. The relation named RSD between the production data of the workshop and the state of the workpiece, the working procedure and the workstation has been represented. The relation named RSI between the state of the information node and the state of the workpiece, the working procedure and the workstation has also been set forth. Thus the relation between the production data and the state of the information has been established. The model of the workshop-production-state has been built accordingly. The model of the workshop-production-state is simplified by adopting the model of the information node.The ideal production state of the workshop-production-field is determined by benchmark production process, and the benchmark production process is formulated by production scheduling. The actual production process runs according to the activation sequence of information nodes which is determined by the benchmark production process. That the actual production process is consistent with the benchmark production process is the controller’s expectation. However, the actual production process deviates from the benchmark production process frequently as there are lots of random factors in the workshop-production-field, and the deviation symboled e(t) between actual production state and ideal production state comes into being. It can be compared one variance with the other only they are homogeneous, so the deviation e(t) was subdivided into the deviation of production time and the deviation of production cost etc. The calculation methods of the deviation e(t) was represented. By introducing the concept of the correlative module of information nodes, the mechanism that the deviation e(t) influences the production process was represented. That mechanism is the basis of expounding prediction deviation.It needs to reschedule frequently to meet the demand of actual production with the changes in the actual production process. To improve the production efficiency of the enterprises, to reduce the production costs and to improve the products’quality are the goals of the dynamic production scheduling. That is the multi-objective optimization problem. The concept of equivalent costs was put forward to convert the multi-objective optimization problem into single-objective optimization problem. On the basis of elaborating the relation among the time, the costs and the quality, they are converted into equivalent costs. It is the objective function of the scheduling problem that minimizes the equivalent costs. And the constraints of the scheduling problem are that the deviations of the production state do not exceed the threshold quantity. Several typical situations that trigger dynamic production scheduling have been represented according to the workshop production state. In connection with the characteristics of these situations, the suitable scheduling policy and scheduling algorithm were put forward to achieve the scheduling objective which minimizes the equivalent costs. The scheduling algorithm must be efficient to meet the real-time nature of the dynamic production scheduling. The scheduling algorithm of phased optimization based on information node was put forward, with the efficient nature by combining heuristic rules and neighborhood-search algorithm that takes the optimal solution in the time neighborhood as the heuristic search direction.On the basis of the research above, the architecture of the remote lean cockpit was built, and the theory and methods represented in this paper are verified to be feasible with a experimental case.更多还原