基于图像流预测和网络化调度的广域运维租赁产线群组维护建模方法研究

As an advanced manufacturing paradigm for saving huge investment, promoting equipment supply, emphasizing core production and integrating outsourcing maintenance, the leasing manufacturing, which extends traditional manufacturing value chain, has been a new trend in the transformation and upgrade of the industry. The maintenance scheduling and adjustment decision-making to serve the lessees is an essential aspect of product-service package provided by the lessor. Previous researches usually focus on single-location maintenance models for one single leased machine or one single production line. In this project, a multi-level lease-oriented maintenance modelling technique is developed. This comprehensive research focuses on the relationships and interactions of the machine-level health prognosis, the system-level opportunistic maintenance and the network-level resource dispatch. In the proposed multi-location lease-oriented maintenance modelling methodology, the degradation image monitoring and failure distribution prediction for each individual leased machine are studied as the support point; the system maintenance opportunity and profit optimization choice for a leased production line are utilized as the research masterstroke; while the maintenance resource scheduling and outsourcing team dispatch for the multi-location leased lines are researched as the optimization surface. This multi-level modelling framework shows the timeliness, dynamic and agility of the lease-oriented maintenance decision-making. This research will lead to a new scientific basis and practical tools for reliability engineering of the leasing manufacturing paradigm. Meanwhile, it has important theoretical significance and practical application value for optimizing the health management of multi-location leased production lines.

作为有效节省产线投资、促进装备供给、聚焦核心生产、统筹外包维护的先进制造模式，延伸传统制造价值链的租赁化生产正逐渐成为制造业转型升级的新趋势。服务多区域承租方的维护规划与调度决策，是出租方提供“产品+服务”装备综合服务中的关键一环。现有研究多是针对单区域租赁设备或产线的狭域运维模型，本研究提出多层级租赁维护建模技术，从设备层健康预测、系统层机会维护、网络层资源调度的相互联系、相互作用中综合开展研究。本项目构建的广域运维租赁维护建模的整体解决方案，以多类型租赁设备的衰退图像监测和失效分布预测为支撑点，以多设备租赁产线的系统维护机会和利润择优规划为研究线，以多工厂租赁网络的维护资源调度和外包团队派遣为优化面，在多层级的建模架构中体现租赁维护决策的实时性、动态性和全局性。本项目的研究，对丰富租赁化生产模式的可靠性工程科学基础和实用工具，及优化广域租赁产线群组健康管理具有重要的理论意义和应用价值。

项目围绕租赁化生产模式中服务多区域承租方的维护规划与调度决策，提出了多层级租赁维护建模技术，从设备层健康预测、系统层机会维护、网络层资源调度的相互联系、相互作用中进行深入研究。本项目主要研究成果：1）通过建立衰退图像流张量驱动预测模型，实现多尺度退化特征提取与多类型失效分布推断，解决设备层多类型租赁设备的衰退图像监测和失效分布预测问题；2）提出利润择优动态性机会维护策略，动态计算机会维护导致的租赁利润增益和损减，解决系统层多设备租赁产线的系统维护机会和利润择优规划问题；3）提出维护团队及资源广域调度决策方法，整合设备衰退更新、系统维护机会和团队路径规划，解决了网络层多工厂租赁网络的维护资源调度和外包团队派遣问题；4）通过燃气轮机叶片加工流水线群组为对象的维护决策与实证分析，验证了本项目的多层级租赁维护建模技术的实时性、动态性和全局性。本项目丰富租赁化生产模式的可靠性工程科学基础和实用工具，解决广域租赁产线群组健康管理的核心问题，为广域运维租赁维护建模提供有效的科学指导。