Towards CyberSHM: autonomous acousto-ultrasonic health monitoring of operational composite structures

对在役安全关键结构进行持续监测，以实时评估其运行状况，这受到了极大的关注，是一个高度热门的研究领域。除其他因素外，这可归因于以下两个因素。i)下一代复杂复合材料结构的快速发展及其作为轻量化结构在航空航天、海上/陆上能源基础设施、汽车和核工业等多个行业的普遍应用。ii)在数据丰富的环境中，自动化的显著进步，以及行业利用其优势转变其传统（通常是干预主义）实践的巨大需求。对下一代复杂复合材料结构的重大研究投资（例如正在进行的EPSRC拨款EP/T011653/1）及其在工业应用中的迅速普及，使这些结构的监测成为关注和挑战。对这些结构对细微的、几乎看不见的损伤（如隐性脱粘、纤维/基质开裂）的敏感性的研究表明，后者会严重危害结构的完整性，并可能导致灾难性的破坏。最近发生在客运航班上的多起灾难性事故，涉及世界上昔日最大的飞机制造商制造的飞机，这理所当然地加强了对此类结构供公众使用的安全性、可使用性和适用性的审查。与此相结合的目标是采用更绿色和可持续的结构（以实现2016年《巴黎气候协定》中承诺的全球排放目标），并在不影响安全性的情况下降低与检查和维护相关的运营成本。与此同时，随着工业4.0中工业物联网的范式转变，无处不在的普适计算与先进的传感和通信技术相结合，开发安全关键工程结构的结构健康监测（SHM）解决方案已经成为一种必要，这些解决方案可以与这种智能、数据丰富的自动化环境并驾齐下，从中获益，并能够无缝适应。该提案旨在对轻质复合材料结构的监测进行基础科学研究和技术实施，以弥合SHM构想的未来愿景与现有评估结构健康的干预主义实践之间的差距。该项目的目标是解决实时声-超声监测（类似于“聆听损伤”和/或结构响应变化）的挑战，并使用多管齐下的方法对运行结构进行在线损伤识别，关键组件是：a)物理驱动的底层模型或各种操作/环境条件下结构超声波导行为的数字等效；同步和利用机载传感网络收集的损伤事件（如工具掉落、分层、裂纹）的结构声指纹，用于数据驱动的损伤事件训练和分类；c)实时损伤识别工具箱(识别位置；类型和严重程度)，这是数据驱动的（原位传感器数据）和模型信息（基于物理的结构波导的理解），以提供早期损伤的量化指标及其估计的置信度。该项目采用了一种新颖的方法，将基于物理的表征结构声学特征与混合被动-主动声学-超声监测数据相结合，并对监测结构进行网络物理监测或在役结构的CyberSHM。

Continuous monitoring of in-service safety-critical structures for real-time assessment of their operational health is receiving significant attention and is a highly topical area of research. This is can be attributed, among others, to the following two factors.i) The rapid evolution of next-generation complex composite structures and their ubiquitous use as lightweight structures in several industries ranging from aerospace to offshore/onshore energy infrastructure, automotive and nuclear industry.ii) The significant advancement of automation within a data-rich environment and the immense appetite of industries to leverage its benefits for transforming their traditional, often interventionist, practices.Significant research investment into next-generation complex composite structures (such as the ongoing EPSRC grant EP/T011653/1) and their rapid uptake in industrial usage has brought to the fore concerns and challenges around monitoring of these structures. Investigations into the susceptibility of these structures to subtle, barely visible damages (like hidden debonding, fibre/matrix-cracking) reveal that the latter can significantly jeopardize the structural integrity and can lead to catastrophic failures. The recent multiple catastrophic accidents in the passenger flights involving aeroplanes manufactured by the world's erstwhile largest planemaker has rightly enhanced the scrutiny on the safety, serviceability and suitability of such structures for public use. This is coupled with objectives for employing greener and sustainable structures (to meet the global emissions target as pledged in the Paris Climate Agreement 2016) and reducing operational costs associated with their inspection and maintenance without compromising on safety.Concurrently, with a paradigmatic shift towards industrial internet of things within Industry4.0 with ubiquitous, pervasive computing coupled with advanced sensing and communication technologies, it has become a necessity to develop structural health monitoring (SHM) solutions of safety-critical engineering structures which are abreast of, can reap the benefits of and are able to fit seamlessly into this intelligent, data-rich environment of automation. The proposal is aimed at fundamental scientific investigation into and the technological implementation of monitoring of lightweight composite structures to bridge the gap between the conceived futuristic vision of SHM and the existing interventionist practices of evaluating structural health. The objective of this project is to address the challenge of real-time acousto-ultrasonic monitoring (akin to "listening for damages" and/or changes in structural response) and online damage identification of operational structures using a multi-pronged approach with the key components being -a) physics-driven underlying model or digital equivalent of structural ultrasonic waveguides behaviour under various operational/ambient conditions,b) the extraction, synchronization and utilization of structural acoustic fingerprints of damage events (such as tool drop, delamination, cracks) as collected by the onboard sensory network for data-driven training and classification of damage events andc) a real-time damage identification toolbox (identifying the location, type and severity) which is both data-driven (in-situ sensor data) and model-informed (physics-based understanding of structural waveguides) to give quantified metrics of incipient damage along with their estimated confidence. The project takes the novel approach of assimilating physics-based characterization structural acoustic characteristics with data from hybrid passive-active acousto-ultrasonic monitoring and interrogation of the monitored structures for a cyberphysical monitoring or CyberSHM of in-service structures.