研究端边云三层数据采集方法、数据传输接口方案、数据存储、数据分析、数据共享融合与治理流程；研究面向多场点多主体的数据处理架构；支持异构海量数据协同的数据治理机制，建立一个高效共享的数据利用环境。面向异构互联装备状态信息个性需求，提出基于通信接口特征的内部状态数据采集方法；研究多源异构数据清洗、转换技术，提升数据准确性和价值密度；基于信息资源服务的数据接入方法，研究多源异构数据接口相关准则，定义对应接口规范、开发接入工具。研制基于5G技术智能终端系统，研发多协议数据采集解析边缘计算软件。构建多源装备数据混合本体异构数据集成、任务引导的装备群组执行过程模型，数据完整性约束验证方法，建立制造和运维数据协同交互和迭代的动态演化理论；建立统一的互操作数据环境，屏蔽底层异构数据源差异，实现不同来源和不同格式的数据再整合。研究面向智能互联装备协同制造运维的安全多方计算框架；研发面向多方安全计算的算法软构件；研究面向安全多方计算的数据传输隐私保护技术，解决多场点多主体的自治区域在协作时私有敏感信息被丢失、被破坏、被篡改问题，保障协同过程中的隐私和安全。This project studies the three-tier data acquisition method,data transmission interface scheme,data storage,data analysis,data sharing fusion and governance process,studies the data processing architecture for multi-site multi-subject,supports the data management mechanism of heterogeneous massive data collaboration,and establishes an efficient and shared data utilization environment.Facing the individual demand of heterogeneous interconnected equipment status information,this paper puts forward the internal state data acquisition method based on the characteristics of communication interface,studies the multi-source heterogeneous data cleaning and transformation technology to improve data accuracy and value density,studies the relevant criteria of multi-source heterogeneous data interface based on the data access method of information resource service,defines the corresponding interface specification and develops the access tool.Develop an intelligent terminal system based on 5G technology,and develop multi-protocol data acquisition and analysis edge computing software.Build a multi-source equipment data hybrid ontological heterogeneous data integration,task-guided equipment group execution process model,data integrity constraint verification method,establish the dynamic evolution theory of collaborative interaction and iteration of manufacturing and operational data,establish a unified interoperability data environment,mask the differences of the underlying heterogeneous data sources,and realize the re-integration of data from different sources and formats.This project also studies the secure multi-party computing framework for the co-manufacturing operation and operation of intelligent and connected equipment,develops the algorithmic software component for multi-party security computing,studies the data transmission privacy protection technology for the security multi-party computing,solves the problem of the loss,destruction and tampering of private sensitive information in multi-site multi-subject autonomous areas when collaborating,and guarantees the privacy and security in the collaborative process.

本课题以地下交通领域复杂装备隧道掘进机为研究对象，旨在研究成套装备群组制造运维互馈协同模型和算法，形成相应软构件；研究效用驱动的多体互馈原语体系，构建制造/运维协同服务集成机制；研究基于脚本驱动的微服务高效动态集成技术，基于容器的微服务高效编排技术，构建开放式制造/运维互馈服务协同软构件集成架构，研发软构件库管理机制；研发制造/运维互馈服务协同集成引擎，构建微服务基础平台，突破制造/运维互馈服务协同集成技术，形成统一应用云服务接口API,解决装备跨域业务交织互馈难的问题。Taking the complex equipment TBM in the field of underground transportation as the research object,this paper aims to study the mutual feedback cooperation model and algorithm between manufacturing and operation of complete equipment group,and form software components.This topic includes the following three sub research points.The first research point is to study the utility-driven multi-body mutual feedback primitive system and build the service integration mechanism based on manufacturing operation and maintenance collaboration.The second research point is to study the script-driven based microservice efficient dynamic integration technology and container based microservice efficient choreography technology,and then build an open software component integration architecture based on mutual feedback between manufacturing and operation and service collaboration,so as to develop the software component library management mechanism.The third research point is to develop a service collaborative integration engine based on mutual feedback between manufacturing and operation,build a basic platform of microservice,break through the service collaborative and integration technology based on mutual feedback between manufacturing and operation,and form a unified application of cloud service interface API,so as to solve the problem of mutual feedback in cross domain business interweaving of equipment.

针对跨域装备互通难、系统应用部署难的问题，分析了云平台业务需求和业务之间的交互关系，基于分布式分层集群理念，进行了平台架构研究，采用HADOOP分布式集群，搭建装备制造/运维数据中心。经过分析目前隧道掘进机制造运维的状况，梳理了隧道掘进机制造协同业务流程、成套装备运维需求及流程、成套运维/群组制造互馈信息及模式，进行了制造/运维集成平台服务应用系统应用场景设计。进行地下交通工程装备制造/运维协同标准大纲及内容的策划，为平台的建设提供了标准支撑。Aiming at the problems of cross domain equipment interoperability and system application deployment,this paper analyzes the business requirements of cloud platform and the interaction relationship between businesses,studies the platform architecture based on the concept of distributed hierarchical cluster,and uses Hadoop distributed cluster to build the equipment manufacturing/operation and maintenance data center.After analyzing the current situation of TBM manufacturing and operation and maintenance,this paper combs the TBM manufacturing collaborative business process,complete equipment operation and maintenance requirements and processes,complete equipment operation and maintenance/group manufacturing mutual feed information and mode,and designs the application scene of manufacturing/operation and maintenance integrated platform service application system.Plan the outline and contents of equipment manufacturing/operation and maintenance coordination standards for underground transportation engineering,which provides standard support for the construction of the platform.

本报告对国内外关于装备服役状态健康评估，姿态智能控制技术和需求预测方法等研究现状进行了综述，明确了采用数据驱动的健康状态评估方法，基于深度强化学习的盾构姿态智能控制技术，和基于时间序列和维修的结合方法。本年度主要完成了七项任务：一是设计了“运行参数—多阶段工况”的跨域数据联合标准化处理技术：包括多源数据特征提取，多阶段工况统计和联合标准化处理；二是实现了运行状态参数敏感性分析：包括低价值参数剔除，非线性相关性分析和单调性计算；三是提出了健康评估综合特征指标定义方法：包括相关系数阈值选择，冗余特征消除和综合特征指标定义。四是基于深度强化学习的盾构姿态智能控制技术及模型构建研究，包括深度神经网络姿态控制模型和基于随机森林的姿态控制模型。同时，进行了盾构姿态实时智能控制案例研究与原型系统的场景规划。五是研究了面向成套装备运行阶段智能预测性维修服务方法，包括装备运行状态时序数据处理，成套装备故障诊断技术和运行状态数据源共享规则。六是建立了备件资源调度系统建设方案。七是构建了备件需求预测的方法模型，其中调度系统主要包括数据、算法模型、应用功能三大模块。数据包括企业的ERP和调度所需的其它数据等。系统主要需要需求预测、库存优化、调度优化三大算法模型支撑。在应用功能层面主要包括技术数据维护、需求预测、库存控制、调度管理四个应用。本年度所完成的任务为后续总体系统搭建及软件开发打好了坚实基础。This report summarizes both the domestic and foreign research status of equipment service status health assessment,attitude intelligent control technology and demand prediction methods,and explicitly adopts the data-driven health status assessment method,shield attitude intelligent control technology based on deep reinforcement learning,and the combination method based on time series and maintenance.Seven tasks were completed this year:(1)the cross domain data joint standardization processing technology of"operation parameters-multi-stage conditions"was designed,including multi-source data feature extraction,multi-stage working condition statistics and joint standardization processing;(2)the sensitivity analysis of operating state parameters was realized,including low value parameter elimination,nonlinear correlation analysis and monotonicity calculation;(3)comprehensive feature index of health assessment definition method was proposed,including correlation coefficient threshold selection,redundant feature elimination and comprehensive feature index definition;(4)the research on shield attitude intelligent control technology and model construction based on deep reinforcement learning,including deep neural network attitude control model and attitude control model based on random forest.At the same time,the case study of real-time intelligent control of shield attitude and the scene planning of prototype system were carried out;(5)the intelligent predictive maintenance service method for the operation stage of complete equipment was studied,including equipment operation state time series data processing,complete equipment fault diagnosis technology and operation state data source sharing rules;(6)the construction scheme of spare parts resource scheduling system was established;(7)a model for the spare parts demand forecasting was built.The scheduling system mainly includes three modules:data,algorithm model and application function.The data includes ERP of the enterprise and other data required for scheduling.The system needs three algorithm models:demand forecasting,inventory optimization and scheduling optimization.At the level of application function,four applications were included,i.e.,technical data maintenance,demand forecasting,inventory control and scheduling management.The tasks completed this year have laid a solid foundation for the subsequent overall system construction and software development.

本报告分别总结了本课题四个任务的年度重要进展情况，具体如下:任务一：已将盾构掘进装备核心部件(主驱动系统为主)的设计制造划分为系统数字化选型设计和系统生产两部分并构建数字化数据库，剖析运维知识与协同设计制造要素间的关系，将系统数字化选型设计和系统生产过程模块化建模并优化，采用系统动力学仿真进行系统验证，有效实现盾构掘进装备核心部件的协同设计制造。任务二：通过对案例库内大量的选型案例数据进行分析和挖掘，正在构建选型规则库，规则库中包含了所有结构型式选择的产生式规则，分析地质条件和开挖要求及参数等与盾构掘进机选型的关系，将其转变为规则。任务三：通过TBM云计算平台实现了一站式技术服务支撑，通过一键式自助服务实现一揽子整体交付，具体成果包括一键式云服务、一站式数据存储平台、一体式数据计算平台、一站式数据分析平台，通过建立掘进设备关键部件相关运行性能参数和工况之间的映射关系，可以提高设计制造的精度，为设备的运行维护提供理论支持。任务四：逐步深入多场点分布式协同制造相关基础理论研究及制造资源云调度平台研发，搭建了多企业协同制造资源模型基本框架，完成分布式协同制造资源建模、分布式企业制造资源能力评价模型、多目标memetic算法求解考虑任务转移的分布式协同制造模型等部分研究内容，为制造企业提供资源共享平台，提升设备利用率。This report summarizes the annual important progress of the four tasks of this subject,as follows:Task 1:The design and manufacturing of TBM core components(mainly the main drive system)has been divided into two parts:system digital type selection design and system manufacturing,and a digital database has been built.The relationship between operation and maintenance knowledge and collaborative design and manufacturing elements has been analyzed,and the system digital type selection design and system manufacturing process have been modeled and optimized,using system dynamics simulation to verify the system,so as to effectively realize the collaborative design and manufacturing of the core components of TBM.Task 2:The type selection rule base is being built based on the analysis and studying of a large number of type selection cases data in the case base.The rule base contains the production rules for the selection of all structural types.The relationships among geological conditions,excavation requirements and parameters and the type selection of TBM is analyzed and transformed into rules.Task 3:Realize one-stop technical service support through TBM cloud computing platform.The realizing of package overall delivery through one click self-service includes the specific achievements include one click cloud service,one-stop data storage platform,integrated data computing platform and one-stop data analysis platform.By establishing the mapping relationship between relevant operation performance parameters and working conditions of TBM key components,the accuracy of design and manufacturing can be improved to provide theoretical support for TBM operation and maintenance.Task 4:Gradually deepen the basic theoretical research related to multi-site distributed collaborative manufacturing and the research and development of manufacturing resource cloud scheduling platform,build the basic framework of multi-enterprise collaborative manufacturing resource model,and complete the distributed collaborative manufacturing resource modeling,distributed enterprise manufacturing resource capability evaluation model,The multi-objective memetic algorithm solves the distributed collaborative manufacturing model considering task transfer,which provides a resource sharing platform for manufacturing enterprises and improves TBM utilization.