The increased focus on predictive maintenance of safety-critical engineering structures requires an onboard structural health monitoring system, which is reliable and robust to provide accurate predictions of health metrics of structures while also being efficient and streamlined to facilitate autonomous data processing and real-time decision-making capabilities. An onboard structural health monitoring system with the capability to continuously monitor and interrogate a structure, describe its current state, and assess the operational risks of the degraded structure needs to be developed and matured so that it can be deployed in practical, real-time monitoring scenarios. This would constitute a cyberphysical system in structural health monitoring. A cyberphysical system is a mechanism that is controlled by computer-based algorithms integrated with the Internet and working with users. There exists a physical domain that is under examination and its digital counterpart, which is informed by data from the physical as well as simulation models. While there exist multiple surveys on the overarching advantages, limitations, and potential of realizing a cyberphysical system, innovation on structural systems, in-line signal processing, and damage event detection in the context of a cyberphysical system, especially from an experimental point of view is still in its infancy. In this work, we implement a versatile cyberphysical framework - CyberSHM using a sparse network of transducers and an edge computing device. Hosted on the structure of interest, the transducers possess the capability to interrogate the structure continuously, periodically, on-demand or autonomously when triggered by damage or an unplanned acoustic event. In addition, the device also possesses efficient on-edge feature extraction and signal classification capabilities, which serve as crucial starting points for further damage analysis and characterization on the digital layer.
A Cyberphysical Structural Health Monitoring Framework for Threshold-Free Active Signal Detection and Classification on the Edge
Lorenzo Morini;
2025-01-01
Abstract
The increased focus on predictive maintenance of safety-critical engineering structures requires an onboard structural health monitoring system, which is reliable and robust to provide accurate predictions of health metrics of structures while also being efficient and streamlined to facilitate autonomous data processing and real-time decision-making capabilities. An onboard structural health monitoring system with the capability to continuously monitor and interrogate a structure, describe its current state, and assess the operational risks of the degraded structure needs to be developed and matured so that it can be deployed in practical, real-time monitoring scenarios. This would constitute a cyberphysical system in structural health monitoring. A cyberphysical system is a mechanism that is controlled by computer-based algorithms integrated with the Internet and working with users. There exists a physical domain that is under examination and its digital counterpart, which is informed by data from the physical as well as simulation models. While there exist multiple surveys on the overarching advantages, limitations, and potential of realizing a cyberphysical system, innovation on structural systems, in-line signal processing, and damage event detection in the context of a cyberphysical system, especially from an experimental point of view is still in its infancy. In this work, we implement a versatile cyberphysical framework - CyberSHM using a sparse network of transducers and an edge computing device. Hosted on the structure of interest, the transducers possess the capability to interrogate the structure continuously, periodically, on-demand or autonomously when triggered by damage or an unplanned acoustic event. In addition, the device also possesses efficient on-edge feature extraction and signal classification capabilities, which serve as crucial starting points for further damage analysis and characterization on the digital layer.
I metadati presenti in IRIS UNICA sono rilasciati con licenza Creative Commons CC0 1.0 Universal, mentre i file delle pubblicazioni sono protetti da diritto d'autore, salvo diversa indicazione.
