差异工件的单机批处理在线优化控制方法

发布时间：2018-01-13 22:32

本文关键词：差异工件的单机批处理在线优化控制方法　出处：《合肥工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：在现代化大规模生产环境下,批量化工件处理可以有效地提高生产效率,从而满足现代化生产需求。因此,研究批处理的优化控制问题,具有重要的理论和现实意义。该问题也是生产优化领域的一个重要研究方向。本文将预测控制机制引入批处理问题中,提出一种传送带给料分批加工模型。在该模型中,加工主体是具有批量化工件处理能力的设备,且系统能够沿着工件到达的方向向前观察一段距离,即前视距离,并能通过传感器获知前视距离内的工件信息。本文在考虑工件尺寸和加工时间存在差异的情形下,重点研究单机模型的在线优化控制问题。首先,以前视距离为控制变量,无穷时段内的工件处理率最大为优化目标,建立了系统的优化模型。该模型包含两个决策优化过程,即工件分批决策和行动选择决策。针对工件分批决策,本文根据系统的运行模式,提出一种以批处理机加工周期内加工能力浪费比最小为准则的工件分批规则；对于行动选择决策,考虑到系统的复杂性,文中引入模型无关的Q学习优化方法,以求解最优前视控制策略。另外,本文还考虑了需求驱动的生产模式,提出一种基于需求驱动的传送带给料分批加工模型,研究了其在线优化控制问题。首先,以无穷时段内的利润损失率最小为优化目标,建立了系统的实时动态分批优化模型。然后,给出了一种基于Q学习的在线优化控制方法。同时,本文进行了相应的仿真分析,结果验证了所构建模型的合理性以及在线优化控制方法的有效性。仿真实验中还分析了不同分批规则以及一些物理参数对系统性能的影响,为系统的设计和优化控制提供了参考。
[Abstract]:In the modern large-scale production environment, batch processing can effectively improve the production efficiency and meet the needs of modern production. Therefore, the optimization control problem of batch processing is studied. It has important theoretical and practical significance. This problem is also an important research direction in the field of production optimization. In this paper, predictive control mechanism is introduced into batch processing. In this model, the processing agent is the equipment with the ability of batch processing, and the system can observe a distance along the direction of the work piece. In this paper, considering the difference of workpiece size and processing time, we focus on the on-line optimization control of single machine model. In this paper, the system optimization model is established, in which the distance is regarded as the control variable and the maximum processing rate of the workpiece in infinite period is considered as the optimization objective. The model consists of two decision-making and optimization processes. In this paper, according to the operation mode of the system, a rule of job batching is proposed, which is based on the minimum ratio of processing capacity waste in the machining cycle of batch processing. Considering the complexity of the system, a model-independent Q-learning optimization method is introduced to solve the optimal forward view control strategy. In addition, the demand-driven production mode is also considered in this paper. This paper presents a demand-driven feeding batch model for conveyors, and studies its on-line optimal control problem. Firstly, the minimum profit loss rate in an infinite period of time is taken as the optimization objective. The real-time dynamic batch optimization model of the system is established. Then, an on-line optimization control method based on Q-learning is presented. At the same time, the corresponding simulation analysis is carried out. The results verify the rationality of the proposed model and the effectiveness of the on-line optimization control method. The effects of different batch rules and some physical parameters on the performance of the system are also analyzed in the simulation experiment. It provides a reference for the design and optimization control of the system.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TB497

【相似文献】