Accurate detection of structural damage is essential for ensuring the safety and reliability of bridges. However, traditional vibration-based approaches often struggle to capture rich feature representations and adequately model spatial dependencies among sensors. This study proposes a novel bridge damage detection framework, LGSTA-GNN, which integrates local–global spatiotemporal learning with graph neural networks. The framework first extracts multi-scale temporal–frequency features using a multi-scale feature extraction module. A local graph feature extraction module then models intrinsic spatial relationships through graph convolutions, while a global graph attention module adaptively captures inter-sensor dependencies by emphasizing structurally informative nodes. A benchmark dataset generated from a scaled bridge model under progressive damage states is used to evaluate the proposed method. Extensive experiments demonstrate that LGSTA-GNN outperforms multiple graph neural network variants and conventional deep learning techniques, achieving superior accuracy, precision, recall, and F1-score. The confusion matrix and t-SNE visualization further verify its enhanced discriminative capability and robustness. Ablation studies confirm the contribution of each module, highlighting the effectiveness of global attention in identifying subtle structural deterioration. Overall, LGSTA-GNN provides an effective and interpretable solution for intelligent bridge damage detection, with strong potential for practical structural health monitoring and real-time safety assessment.

Real-time monitoring of surface cracks in reinforced concrete (RC) beams is critical to structural safety and service performance evaluation. Current structural crack monitoring still faces prominent scientific and technical bottlenecks: conventional unidirectional sensors cannot achieve multi-directional collaborative sensing, rigid piezoelectric materials exhibit poor compatibility with the large deformation of concrete, and there is a lack of quantitative mapping relationships from sensing signals to crack parameters, making it difficult to simultaneously measure crack width, angle, and morphology. This paper presents a novel ring-shaped piezoelectric sensor based on polyvinylidene fluoride (PVDF) and an annular piezoelectric sensing mechanism for real-time monitoring of crack angle, width, and morphology. The sensor incorporates a laminated structure with four strip sensing units for multi-directional strain detection. Experiments were conducted on RC beams under various loading conditions, and finite element analysis was performed using COMSOL Multiphysics. An innovative crack damage index (B) was introduced to assess structural damage quantitatively. Results demonstrate high sensor sensitivity and stable output. Voltage signals increase both with crack width and crack angle, showing responses of 0.045 mV, 0.041 mV, and 0.023 mV for crack angles of 60°, 45°, and 30°, respectively, at a crack width of 9 mm. Strong consistency between experimental and simulation data validates the effectiveness of the mechanism in monitoring the direction, width, and types of cracks. The crack damage index B exhibits a positive correlation with the structural stress response, enabling a quantitative assessment of damage. This study is applicable to the prestressed concrete box girders and T-beams commonly used in large-span bridges, which are typically with a main span of 20–50 m, a beam length of 6–30 m, a section height of 1.2–2.5 m, and designed for Grade C35–C50 concrete. The findings provide a practical foundation for real-time crack monitoring in large-scale bridge beam members.

Long-term data loss resulting from sensor malfunctions, communication interruptions, and other factors in Structural Health Monitoring (SHM) significantly undermines the reliability of damage identification and safety assessment. Existing methods—ranging from statistical approaches and low-rank matrix completion to traditional machine learning and deep learning imputation techniques—often suffer from either limited accuracy or excessive model size and slow inference, making deployment in resource-constrained scenarios difficult. To address these challenges, this paper proposes TEFN–Imputation, a lightweight and efficient time-series imputation model. This model utilizes observation-driven non-stationary normalization to mitigate the impact of time-varying characteristics and dimensional discrepancies. It employs linear projection for temporal length alignment and constructs BPA-style mass representations from dual perspectives of time and channel. Furthermore, it replaces strict Dempster–Shafer belief combination with an expectation-based evidential aggregation (readout), thereby significantly reducing computational overhead while enabling uncertainty-aware evidential indicators for interpretation rather than claiming a direct accuracy gain from uncertainty modeling. The observed accuracy and robustness improvements are primarily attributed to the normalization and dual temporal–channel modeling design under the same lightweight readout. Systematic experiments on two real-world bridge monitoring datasets, Z24 and Hell Bridge, demonstrate that TEFN consistently maintains low Mean Absolute Error (MAE) and minimal volatility across various combinations of training and testing missing rates, exhibiting high robustness against variations in missing rates and train–test mismatches. Concurrently, compared to RNN and large-scale Transformer baselines, TEFN reduces parameter count and CPU inference time by one to two orders of magnitude. Thus, it achieves a superior trade-off among accuracy, efficiency, and model scale, making it highly suitable for online SHM and imputation tasks in practical engineering applications. Across the settings on Z24, TEFN achieves a mean MAE of 0.218 with a standard deviation of 0.002, while using only 0.02 MB parameters and 2.73 ms per batch CPU inference.

Steel pipelines for oil and gas transportation serve as the lifeline of energy conveyance, and their long-term safe operation constitutes a crucial safeguard for energy security. Nevertheless, in complex service environments, local defects formed on the inner pipe wall due to medium corrosion have emerged as a prominent hidden danger endangering pipeline integrity. Accurate evaluation of the residual strength of pipelines with corrosion defects is not only the technical foundation for ensuring the safe operation of pipelines, but also the key basis for formulating scientific maintenance strategies and prolonging the service life of pipelines. Taking three grades of steel pipelines (X52, X65 and X80), which represent the typical strength grades commonly used in long-distance oil and gas transmission pipelines, as the research objects, this paper establishes a three-dimensional finite element model of single-point uniform corrosion defects considering the nonlinear material behavior, and systematically investigates the influence laws of geometric parameters (depth, length and width) of corrosion defects on the failure pressure of pipelines under the action of monotonic internal pressure load. The accuracy of the proposed finite element model is verified by comparison with the test data from thirteen groups of full-scale burst experiments. On the basis of parametric analysis results, an explicit and high-precision predictive model for failure pressure is developed. The research findings reveal that corrosion depth acts as the dominant factor affecting pipeline failure pressure with a distinctly nonlinear influence characteristic: the load-bearing capacity of pipelines drops drastically when the relative depth d/t exceeds 0.6, where d is the corrosion depth and t is the pipe wall thickness. There exists a critical value for the impact of corrosion length, beyond which its weakening effect on failure pressure tends to level off. Within the commonly encountered engineering range (20~100°), corrosion width exerts a negligible influence on pipeline failure pressure and thus can be overlooked in engineering evaluation. In comparison with conventional industry assessment methods such as ASME B31G, DNV RP-F101, PCORRC and SY/T 6151, the newly established predictive model features higher prediction accuracy and broader applicability, which provides on-site engineers with a powerful theoretical tool and practical formula for the rapid and accurate evaluation of the residual strength of corroded pipelines.