2026-10-10T16:52:35Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33957734 2026-09-21T17:47:55Z category_734 category_931 category_811 portal_5 item_type_3 month_year_09_2026

<p>Damage cases applied to the My Thuan bridge FEM.</p> Quan Pham Hong (25084847) Trung Vu Manh (25084850) Bich Nguyen Thach (25084853) Le Nguyen Dan (25084856) Hoa Tran Ngoc (25084859) Biological Sciences not elsewhere classified Information Systems not elsewhere classified Mathematical Sciences not elsewhere classified wavelet gating module two benchmark datasets streaming acceleration windows shown strong potential resolution filtering within physically interpretable sub moderate computational cost localised events confined identifying subtle changes finite element model filtered signal via existing architectures share deep learning models 4 ablation variants purely convolutional backbone proposed architecture operates highest mean accuracy frequency component occurs discrete wavelet transform 10 baseline models specific frequency bands paper proposes spectraltcn making spectraltcn suitable dwt simultaneously preserves div >< p z24 bridge benchmark strongest baseline stayed bridge proposed approach global frequency generalized mean frequency information domain transform convolutional block bands corresponding unlike fft structure ’ spectraltcn attains spectral location serious drawback performs learnable key limitation inverse dwt induced anomalies generalisation capability fold cross dynamic response discards information decomposition level <div><p>Vibration-based structural health monitoring (SHM) detects damage by identifying subtle changes in a structure’s dynamic response. Deep learning models have shown strong potential for automating this classification task. However, most existing architectures share a key limitation: they either process signals purely in the time domain, or apply a global frequency-domain transform that discards information about when each frequency component occurs. This is a serious drawback, because structural damage typically manifests as short, localised events confined to specific frequency bands. This paper proposes SpectralTCN, a temporal convolutional network augmented with a Wavelet Gating Module (WGM) that performs learnable, data-dependent multi-resolution filtering within each convolutional block. The WGM decomposes intermediate features via the Discrete Wavelet Transform (DWT) into physically interpretable sub-bands corresponding to distinct structural vibration modes, applies input-adaptive sigmoid gates independently at each decomposition level, and reconstructs the filtered signal via the Inverse DWT (IDWT) with a learnable residual connection initialised to zero. Unlike FFT-based approaches, the DWT simultaneously preserves both time and frequency information, enabling the network to detect both the timing and the spectral location of damage-induced anomalies. Combined with Generalized Mean (GeM) pooling and large-kernel causal depthwise convolutions, SpectralTCN is evaluated on two benchmark datasets: the Z24 Bridge benchmark and a finite element model (FEM)-derived dataset of the My Thuan cable-stayed bridge. Experiments against 10 baseline models and 4 ablation variants, evaluated via stratified 5-fold cross-validation, demonstrate the effectiveness and generalisation capability of the proposed approach: SpectralTCN attains the highest mean accuracy on both benchmarks (92.2% on Z24 and 92.1% on My Thuan) and outperforms the strongest baseline in the 5-fold cross-validation protocol. In addition, the proposed architecture operates on short, streaming acceleration windows with a purely convolutional backbone of moderate computational cost, making SpectralTCN suitable for <b>near-real-time, online damage detection</b> in continuous bridge monitoring.</p></div> 2026-09-21T17:47:24Z Dataset Dataset 10.1371/journal.pone.0358224.t005 https://figshare.com/articles/dataset/_p_Damage_cases_applied_to_the_My_Thuan_bridge_FEM_p_/33957734 CC BY 4.0