Nowadays, drawing up plans to control and manage infrastructural assets has become one of the most important challenges in most developed countries. The latter must cope with issues relating to the aging of their infrastructures, which are getting towards the end of their useful life. This study proposes an automatic approach for tunnel defects classification. Starting from non-destructive investigations using Ground Penetrating Radar (GPR), the deep convolutional neural networks (CNNs), with and without the application of bidimensional Fourier Transform (2D FT), have allowed the classification of several structural defects (e.g., crack, voids, anomaly, etc.) with high accuracy. The proposed methodology eliminates the need for human interpretation of Ground Penetrating Radar profiles and the use of integrative investigations (e.g., video-endoscopy, core drilling, jacking, and pull-out testing) for defects classification. As a result, it has significant speed and reliability that make it both time and cost-efficient.

Ground Penetrating Radar Fourier Pre-processing for Deep Learning Tunnel Defects’ Automated Classification / Marasco, Giulia; Rosso, Marco M.; Aiello, Salvatore; Aloisio, Angelo; Cirrincione, Giansalvo; Chiaia, Bernardino; Marano, Giuseppe C.. - 1600:(2022), pp. 165-176. (Intervento presentato al convegno 23th International Conference on Engineering Applications of Neural Networks (Engineering Applications and Advances of Artificial Intelligence) tenutosi a Crete, Greece nel 17 – 20 June, 2022) [10.1007/978-3-031-08223-8_14].

Ground Penetrating Radar Fourier Pre-processing for Deep Learning Tunnel Defects’ Automated Classification

Giulia Marasco;Marco M. Rosso;Salvatore Aiello;Giansalvo Cirrincione;Bernardino Chiaia;Giuseppe C. Marano
2022

Abstract

Nowadays, drawing up plans to control and manage infrastructural assets has become one of the most important challenges in most developed countries. The latter must cope with issues relating to the aging of their infrastructures, which are getting towards the end of their useful life. This study proposes an automatic approach for tunnel defects classification. Starting from non-destructive investigations using Ground Penetrating Radar (GPR), the deep convolutional neural networks (CNNs), with and without the application of bidimensional Fourier Transform (2D FT), have allowed the classification of several structural defects (e.g., crack, voids, anomaly, etc.) with high accuracy. The proposed methodology eliminates the need for human interpretation of Ground Penetrating Radar profiles and the use of integrative investigations (e.g., video-endoscopy, core drilling, jacking, and pull-out testing) for defects classification. As a result, it has significant speed and reliability that make it both time and cost-efficient.
2022
978-3-031-08222-1
978-3-031-08223-8
File in questo prodotto:
File Dimensione Formato  
eann2022_paper_25.pdf

Open Access dal 11/06/2023

Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 577.41 kB
Formato Adobe PDF
577.41 kB Adobe PDF Visualizza/Apri
Marasco-Rosso-etal_EANN2022.pdf

non disponibili

Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 1.53 MB
Formato Adobe PDF
1.53 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

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/11583/2971530