制造物联网环境下混流制造过程自适应调度方法研究

发布时间：2018-06-25 13:58

本文选题：自适应调度 + 混流制造　；参考：《广东工业大学》2013年博士论文

【摘要】：混流制造是一种以客户需求为导向、在当前大批量定制生产中的常见生产组织模式。其生产订单存在多品种、周期性、数量多变等特点,但生产过程资源需求模型相对固定。混流制造过程不可避免存在诸多不确定的动态事件,如设备故障、紧急插单、质量事故等,导致生产过程无法遵循预定义的基准计划执行。因此对混流制造采取合理的动态调度机制,以消除动态事件对计划执行的影响,保持制造过程的稳定性,成为一个重要科学问题。然而,目前对动态调度的研究多集中在一个理想模型下的调度理论研究,其随机事件的加入也大多基于某种理论模型,没有考虑车间信息反馈断层的问题,缺乏在实际应用中的技术支撑环境,在实际应用中无法应对车间现场的瞬息万变。随着物联网技术飞速发展,实时制造环境下的自适应调度,具备了实现的技术基础。本论文在国家基金“基于RFID的分时段双层实时动态OKP调度理论模型与算法研究(61074146)”支持下,基于制造物联网实时制造环境,对混流制造过程在实时反馈条件下的自适应调度方法进行了研究。论文的具体研究内容如下： 1)、针对目前我国中小企业制造车间存在的信息断层问题,构建了一个基于制造物联网技术的实时制造系统的框架,为后续混流过程的自适应调度研究提供了技术基础。在此框架下,采用了统一接口构建了RFID中间件实现了Multi-Agent封装模式下RFID对象的即插即用接入。然后设计了一个基于RFID-Bus的实时消息处理模式,实现了实时制造消息的处理和反馈统一机制。基于此机制,构建了一个两级RFID-Bus的实时制造系统环境及其在企业制造车间的部署方法。 2)、针对混流制造系统规模庞大,难以求解的难题,提出了一个基于ROPN的制造系统MPN建模机制,给出了具体的建模流程和方法,缩小了混流制造系统模型规模。论文先分析了混流制造过程带转运／缓存约束的非等价并联机制造的特点,通过将等价并行机建模成一个资源提供节点,提出了一种基于制造系统资源建模方法和建模流程。定义了MPN (Manufacturing Petri Net)模型,在模型中阐述了基于实时制造物联网中智能令牌在模型中映射的知识函数,及路径查找办法。 3)、构建了混流制造系统MPN模型后,对MPN中非等价并联机调度问题进行了分析,针对混流模型中并联机调度难题,构建了一个通过调度变迁触发顺序来进行作业调度的离线调度方案,并给出了基于MMAS的蚁群算法的优化方法,并采用田口实验设计方法对算法中的参数最优配置进行了探求,形成了制造系统执行的基准计划。 4)、针对基准计划在MPN中的执行,首先分析了实时制造环境下大规模定制混流制造过程的动态事件的特点,并将动态事件在MPN中统一映射为资源提供能力的失能事件。根据实时制造环境的特点,采用修正式策略,构建了一个计划与执行交互的在线自适应调度的框架。在这个框架中,通过树结构的决策单元,实时监控制造系统MPN中各作业的执行偏离度,根据偏离度分布情况进行实时修正。然后,根据决策树结构特点,提出了一个基于层级反馈的在线调度方法,并对在线调度方法进行了实现。 5)、根据论文提出的自适应调度框架和理论,开发了一套自适应调度仿真系统,对制造系统进行了仿真和推演,对算法有效性进行了验证。案例测试结果表明,本论文提出基于制造物联网实时制造环境的自适应调度方法,在某些动态事件模式下,可以有效的消除其对制造系统的影响,提高车间生产效率,验证了论文工作的合理性。本论文研究内容还有许多不足之处需要不断完善和改进,有待今后进一步研究。
[Abstract]:Mixed flow manufacturing is a kind of common production organization pattern which is guided by customer ' s demand . It has many characteristics such as many varieties , periodicity , quantity and so on , but the production process resources demand model is relatively fixed . There are many uncertain dynamic events such as equipment failure , emergency plug - in , quality accident and so on , which can lead to the production process not to follow predefined benchmark plan execution . Therefore , the process of mixed flow is reasonably dynamic scheduling mechanism to eliminate the effect of dynamic event on the plan execution , and maintain the stability of manufacturing process and become an important scientific problem .

However , the researches on dynamic scheduling are mostly focused on the scheduling theory under an ideal model , and the random events are mostly based on a certain theoretical model , and the problems of the workshop information feedback faults are not taken into consideration , and the technology support environment in the practical application is not taken into account .

Based on the real - time manufacturing environment of Internet of manufacture , this paper studied the adaptive scheduling method under real - time feedback condition based on the real - time manufacturing environment .

The contents of the thesis are as follows :

1 ) In order to solve the problem of information faults existing in the manufacturing workshop of small and medium - sized enterprises in China , a frame of real - time manufacturing system based on Internet of manufacture technology is constructed , which provides a technical basis for the adaptive scheduling of the subsequent mixed - flow process .

In this paper , an MPN modeling mechanism based on ROPN is proposed to reduce the size of the mixed - stream manufacturing system .

3 ) After constructing the MPN model of the mixed - stream manufacturing system , the problem of non - equivalent and online scheduling of MPN is analyzed . According to the problem of online scheduling in the mixed - flow model , an off - line scheduling scheme for job scheduling is constructed by scheduling the transition triggering sequence , and an optimization method of the ant colony algorithm based on MMAS is constructed , and the optimal allocation of parameters in the algorithm is explored by using the field - port experiment design method , and a reference plan for the execution of the manufacturing system is formed .

4 ) According to the implementation of the benchmark plan in MPN , firstly , the characteristics of the dynamic event of mass customization mixed - stream manufacturing process in real - time manufacturing environment are analyzed , and the dynamic events are uniformly mapped into the power - loss events of resources in MPN . According to the characteristics of real - time manufacturing environment , a plan and execution interaction - based on - line adaptive scheduling framework is built . In this framework , a hierarchical feedback based on - line scheduling method is proposed based on the characteristics of decision tree structure , and the on - line scheduling method is realized .

5 ) According to the adaptive scheduling framework and theory proposed by the paper , an adaptive scheduling simulation system is developed , and the simulation and deduction of the manufacturing system are carried out , and the validity of the algorithm is verified .

The case test results show that the self - adaptive scheduling method based on the real - time manufacturing environment of the manufacturing internet of things can effectively eliminate the influence on the manufacturing system under certain dynamic event modes , improve the production efficiency of the workshop and verify the rationality of the work of the paper .

There are many deficiencies that need to be perfected and improved , and further research is to be done in the future .
【学位授予单位】：广东工业大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TH186

【参考文献】