Condition Monitoring of Cable Structures using Digital Twin

电缆结构广泛用于各种运输系统，例如：用于铁路电气化的架空接触线（ocl），缆车和桥梁中各种电缆的应用。与传统结构相比，电缆结构的状态监测需要考虑不同的挑战：(1)电缆对腐蚀非常敏感，可能会意外失效；（ii）缆索的固有振动频率与结构的“刚性”构件（例如桥面）有显著差异；（iii）电缆和电缆结构具有复杂和弯曲的几何形状；（iv）在某些情况下，缆索会受到移动的负载（缆车及电缆缆索）的影响。本项目旨在开发一种新的方法来监测这些电缆结构的状态。这将包括对结构进行明智的仪器检测，并结合使用基于系统振动行为的物理模型训练的机器学习模型——一个“数字双胞胎”。基于物理的模型将使用现场测试数据进行校准和验证。它将用于使用在役数据和具有参数变化的模拟来构建快速查找数字孪生。数字孪生将被设计用于实时状态监测；它将用于快速自动检测和诊断结构中正在发生的故障，并启动安全程序（例如关闭协议或标记该区域以进行紧急维护）。特定故障发生的模式可用于通知定期维护指导和计划。该项目的关键研究领域是：(i)传感器设置的优化-所使用的传感技术及其位置都需要仔细选择并集成到诊断模型中。（ii）为基于电缆结构的数字孪生开发物理支持的人工智能技术，该技术具有广泛的固有频率和代表性参数变化。（iii）开发诊断方法，实时识别异常振动行为，在安全危急情况下迅速启动适当的干预措施。

Cable structures are used in a wide variety of transport systems - some examples are:overhead contact lines (OCLs) for electrifying railways, cable-cars and various applicationsof cables in bridges. There are different challenges to consider for condition monitoring ofcable structures compared to conventional structures: (i) cables are highly sensitive tocorrosion and can fail unexpectedly; (ii) natural frequencies of vibration of cables differsignificantly to the 'rigid' elements of the structure (e.g the bridge deck); (iii) cables and cablestructures have complex and curved geometries; (iv) in some cases cables are subject tomoving loads (cable-cars and OCLS).This project aims to develop a novel approach to condition monitoring of these cablestructures. This will consist of judicious instrumentation of the structure in combination with amachine learning model trained using a physics-based model of the systems vibrationbehaviour - a 'digital-twin'. A physics-based model will be calibrated and validated using fieldtest data. It will be used to build a fast-lookup digital twin using both in-service data andsimulations with parametric variations. The digital-twin will be designed for real-timecondition monitoring; it will be used to rapidly and automatically detect and diagnosedeveloping faults in the structure and initiate safety procedures (e.g. shut down protocol orflag the area for urgent maintenance). Patterns in the occurrence of specific faults can beused to inform regular maintenance guidance and planning.The key research areas of the project are:(i) optimization of sensor set-ups - both the sensing technologies used and their placementswill need to be carefully chosen and integrated into the diagnostic models.(ii) development of physics-enabled AI technologies for digital twins of cable-basedstructures, effective with a wide range of natural frequencies and representative parametricvariations.(iii) developing diagnostic methods to identify abnormal vibration behaviour in real-time,enabling rapid initiation of the appropriate intervention in safety-critical situations.