PSI - Issue 84

Available online at www.sciencedirect.com

ScienceDirect

Procedia Structural Integrity 84 (2026) 280–287

© 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 the scientific committee of the Conference Abstract The deployment of dense sensor networks and long-term vibration monitoring in structural health monitoring (SHM) systems has intensified challenges related to data storage, transmission, and processing. Data compression offers a practical solution to mitigate these limitations; however, achieving an appropriate trade-off between compression ratio and reconstruction accuracy for reliable dynamic identification remains challenging. This paper presents a lightweight residual convolutional autoencoder for vibration signal compression and reconstruction, designed to preserve modal characteristics relevant to SHM while maintaining high time domain reconstruction accuracy. The proposed framework employs a joint time–frequency reconstruction loss that balances waveform fidelity and spectral consistency, without relying on predefined sparsity assumptions or prior modal knowledge. The method is evaluated using a synthetic vibration dataset and a real monitoring deployment on the San Jerónimo bell tower in Granada, Spain. Under controlled conditions, accurate reconstruction and modal preservation are achieved at compression ratios up to CR = 8. For the real case, a more conservative compression ratio (CR = 2) is required to ensure stable modal identification, reflecting the influence of environmental variability and measurement noise. Overall, the results demonstrate that the proposed approach effectively reduces vibration data volumes while retaining essential modal information for SHM applications. III Fabre Conference: Existing Bridges, Viaducts, and Tunnels: Research, Innovation, and Applications Preliminary Evaluation of Neural Network–Based Vibration Compression for Modal Preservation in Structural Health Monitoring Andy Duarte-Taño ᵃ *, Rafael Castro-Triguero ᵇ , Rafael Gallego-Sevilla ᵃ , Enrique García Macías ᵃ ᵃ Department of Structural Mechanics and Hydraulic Engineering University of Granada, Campus de Fuentenueva s/n, 18071 Granada, Spain ᵇ Department of Continuum Mechanics and Theory of Structures University of Cordoba, Escuela Politécnica Superior, C/ María, Virgen y Madre, s/n. 14071 Cordoba, Spain

* Corresponding author. E-mail address: andyduarte0192@ugr.es

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 the scientific committee of the Conference 10.1016/j.prostr.2026.06.037