The management of the aging built infrastructure stands as a paramount concern on the political agendas worldwide, bearing far-reaching socio-economic impacts. The growing trend of tragic collapses in recent years underscore the urgent need for efficient structural health maintenance (SHM) strategies to support decision-making in prioritizing intervention and rehabilitation actions. Vibration-based SHM systems utilizing Operational Modal Analysis (OMA) have gained popularity owing to their non-destructive nature, global damage assessment capabilities, and relatively straightforward automation with minimal intrusiveness. Nevertheless, state-of-the-art OMA techniques often face significant scalability limitations, primarily driven by extensive computational requirements and need for substantial expert involvement. In this context, recent advances in the realm of artificial intelligence (AI) offer great promise in addressing these scalability issues, paving the way for next-generation SHM systems. In this light, this work introduces a novel Multitask Learning Deep Neural Network (MTL-DNN) model designed for fast and automated blind source modal identification of structures. By encapsulating the principles of second-order blind source identification (SOBI) within the network's architecture, the proposed model can extract the complex-valued modal components concealed within input raw response acceleration data. This enables the direct extraction of complex-valued mode shapes from the weights of the network, and the corresponding resonant frequencies and damping ratios are estimated through a computationally light single-degree-of-freedom identification algorithm. The efficacy of the presented approach is validated through three case studies: a theoretical non-proportionally damped system, a laboratory steel frame structure, and a real-world reinforced concrete arch bridge. The presented results demonstrate the capability of the proposed technique to conduct near-instantaneous automated modal identification with minimal expert intervention, holding great potential as a scalable technique for SHM of large infrastructural systems.

AI-driven blind source separation for fast operational modal analysis of structures

Hernandez-Gonzalez I. A.;Costante G.;Ubertini F.
2024

Abstract

The management of the aging built infrastructure stands as a paramount concern on the political agendas worldwide, bearing far-reaching socio-economic impacts. The growing trend of tragic collapses in recent years underscore the urgent need for efficient structural health maintenance (SHM) strategies to support decision-making in prioritizing intervention and rehabilitation actions. Vibration-based SHM systems utilizing Operational Modal Analysis (OMA) have gained popularity owing to their non-destructive nature, global damage assessment capabilities, and relatively straightforward automation with minimal intrusiveness. Nevertheless, state-of-the-art OMA techniques often face significant scalability limitations, primarily driven by extensive computational requirements and need for substantial expert involvement. In this context, recent advances in the realm of artificial intelligence (AI) offer great promise in addressing these scalability issues, paving the way for next-generation SHM systems. In this light, this work introduces a novel Multitask Learning Deep Neural Network (MTL-DNN) model designed for fast and automated blind source modal identification of structures. By encapsulating the principles of second-order blind source identification (SOBI) within the network's architecture, the proposed model can extract the complex-valued modal components concealed within input raw response acceleration data. This enables the direct extraction of complex-valued mode shapes from the weights of the network, and the corresponding resonant frequencies and damping ratios are estimated through a computationally light single-degree-of-freedom identification algorithm. The efficacy of the presented approach is validated through three case studies: a theoretical non-proportionally damped system, a laboratory steel frame structure, and a real-world reinforced concrete arch bridge. The presented results demonstrate the capability of the proposed technique to conduct near-instantaneous automated modal identification with minimal expert intervention, holding great potential as a scalable technique for SHM of large infrastructural systems.
2024
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11391/1576802
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? 1
social impact