Abstract: In addressing the monitoring and control issues within intelligent manufacturing systems,a bidirectional digital twin management approach tailored for industry was introduced.The forward aspect of this approach involved creating twin entities through data mapping to offer monitoring services,while the reverse aspect employed simulation-based control optimization to enhance the behavior of physical entities,which achieved a fully closed-loop control over the manufacturing process.By integrating real-world physical data and virtual twin data within a cyber-physical system,a multi-layer architecture was established.A multi-scale and multi-level twin modeling method was devised,and coupling model-based definitions and finite state machine techniques were integrated to construct twin scenarios of physical attributes and behavioral actions using the Unreal Engine.By amalgamating artificial intelligence with behavior simulation models,the contextual data was incorporated into functional services,so that the system could effectively harness fused data,analyze and evaluate equipment health status and generate simulated behavioral controls for the manufacturing process.Finally,a platform was developed for a component manufacturing system to validate the maturity of the proposed model and the reliability of the twin technology.

Key words: intelligent manufacturing system, digital twins, prognostication monitoring, retrospective control

摘要： 针对智能制造系统的监测评估与控制优化问题,提出面向工业现场的正反向数字孪生管理方法。正向由数据映射孪生实体提供监控服务,反向模拟控制优化物理实体行为完成反馈,实现制造过程的全闭环管控。方法由信息物理系统串联物理真实数据与孪生虚拟数据,搭建与五维模型呼应的多层架构。设计了多尺度多层次的孪生建模方法,结合模型定义与有限状态机技术,在虚幻引擎上搭建物理属性和动作行为的孪生场景。通过融入人工智能与行为模拟模型,将上下文数据信息纳入功能服务,使系统能够有效利用融合数据,分析评估设备健康状态,生成仿真行为控制加工过程。最后,面向某装配件智能制造系统搭建了平台,验证了模型成熟度与孪生可靠性。

关键词: 智能制造系统, 数字孪生, 状态监测预估, 反向控制