PSI - Issue 82

Available online at www.sciencedirect.com Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2026) 000–000 Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2026) 000–000

www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia

ScienceDirect

Procedia Structural Integrity 82 (2026) 51–57

© 2026 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of ICSID organizers Abstract This study evaluates the effectiveness of Ground Penetrating Radar (GPR) and Impact Echo (IE) for detecting delamination in a reinforced concrete bridge deck in Grand Forks, ND. Minimally processed signals were transformed using Wavelet with FFT or STFT to create 2D scalograms and classified using ensemble learning. IE achieved 85% accuracy, slightly outperforming GPR at 82%. The Wavelet + FFT combination provided better detection balance, and fusing both modalities improved specificity without reducing accuracy. Findings highlight the promise of multimodal NDE signal fusion for more robust bridge health monitoring. © 2026 The Authors. Copy from the contract: Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of ICSID organizers Keywords: GPR signals; impact echo signals; Wavelet; STFT; FFT; ensemble learning; feature extraction 1. Introduction Bridge health conditions in the past have been evaluated using traditional techniques such as physical inspection, visual inspection, coring, and chain-dragging. In recent times, the use of Non-Destructive Evaluation/Testing (NDE/NDT) approaches has been increasing due to their numerous advantages over traditional methods. Impact Echo (IE), Ground Penetrating Radar (GPR) signals, and thermal images are considered the main techniques for determining defect areas in bridges. 8th International Conference on Structural Integrity and Durability (ICSID2025) Comparative study of GPR and impact echo signals for defect detection using wavelet transform, FFT, STFT, and ensemble learning Faezeh Jafari a, *, Sattar Dorafshan a a University of North Dakota , GrandForks, North Dakota, USA Abstract This study evaluates the effectiveness of Ground Penetrating Radar (GPR) and Impact Echo (IE) for detecting delamination in a reinforced concrete bridge deck in Grand Forks, ND. Minimally processed signals were transformed using Wavelet with FFT or STFT to create 2D scalograms and classified using ensemble learning. IE achieved 85% accuracy, slightly outperforming GPR at 82%. The Wavelet + FFT combination provided better detection balance, and fusing both modalities improved specificity without reducing accuracy. Findings highlight the promise of multimodal NDE signal fusion for more robust bridge health monitoring. © 2026 The Authors. Copy from the contract: Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of ICSID organizers Keywords: GPR signals; impact echo signals; Wavelet; STFT; FFT; ensemble learning; feature extraction 1. Introduction Bridge health conditions in the past have been evaluated using traditional techniques such as physical inspection, visual inspection, coring, and chain-dragging. In recent times, the use of Non-Destructive Evaluation/Testing (NDE/NDT) approaches has been increasing due to their numerous advantages over traditional methods. Impact Echo (IE), Ground Penetrating Radar (GPR) signals, and thermal images are considered the main techniques for determining defect areas in bridges. 8th International Conference on Structural Integrity and Durability (ICSID2025) Comparative study of GPR and impact echo signals for defect detection using wavelet transform, FFT, STFT, and ensemble learning Faezeh Jafari a, *, Sattar Dorafshan a a University of North Dakota , GrandForks, North Dakota, USA

* Corresponding author. Tel.: 3109024934. E-mail address: Faezeh.jafari@und.edu * Corresponding author. Tel.: 3109024934. E-mail address: Faezeh.jafari@und.edu

2452-3216 © 2026 The Authors. Copy from the contract: Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of ICSID organizers 2452-3216 © 2026 The Authors. Copy from the contract: Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of ICSID organizers

2452-3216 © 2026 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of ICSID organizers 10.1016/j.prostr.2026.04.009