Lab-to-field integration in bridge monitoring: a hybrid structural health monitoring framework employing deep learning and unmanned aerial vehicle imagery

The changing landscape of concrete bridge infrastructure health monitoring and informed maintenance strategies: A decade of developments and trends

GraphDARTS: Contrastive graph metric learning for unsupervised differentiable architecture search in Structural Health Monitoring

DamSegment: A curriculum-structured image dataset for damage assessment of concrete dams

Deep generative models in condition and structural health monitoring: Opportunities, limitations and future outlook

Time-Vertex machine learning for optimal sensor placement in temporal graph signals: Applications in structural health monitoring

Validating sideband peak count-index (SPC-I) technique as a hybrid linear/nonlinear ultrasonic technique through numerical modeling and experiment.

Efficient maintenance of infrastructure relies on monitoring and assessment of its condition. New technologies and methods thereby enable a deeper understanding of the materials used and of the structures built. Coda Wave Interferometry (CWI) is currently explored for continuous monitoring of reinforced concrete structures as well as material testing. This ultrasound-based method is sensitive to even small material alterations and therefore suitable for the detection of initial damage stages. Herein, a step-by-step procedure for the evaluation of ultrasonic signals with CWI methods is presented. The described procedure is proposed for ultrasonic signals collected with embedded ultrasonic transducers with a center frequency of 50 kHz to 70 kHz from prism-shaped concrete specimen with dimensions of 400 mm x 100 mm x 100 mm. The raw ultrasonic signal, preprocessing and CWI analysis are described and influences of parameters within the analysis are discussed. The presented procedure allows systematic and comparable analysis of ultrasonic signals generated with similar conditions and therefore contributes to the application of CWI methods for structural health monitoring and material testing.

Developing a condition assessment method for the structural health monitoring (SHM) of unreinforced masonry (URM) structures is an utmost need to identify the remaining service life of the structure, the application of retrofitting techniques, and vulnerability assessment. However, existing methods have focused on a few parameters and not considered the significance and correlation between them. Existing methods for the condition assessment of URM structures use linguistic definitions to classify the degree of distress and lead to ambiguity and human perspective error. The SHM tool developed for this study suggested a numerical framework for evaluating and comparing the structural health of various URM structures. In this study, a new condition assessment method is developed using the fuzzy analytic hierarchy process (FAHP), which incorporates the visual inspections, crack details, structural characteristics, ground characteristics, casual characteristics, and masonry tests. Numerical thresholds were incorporated into this model to mitigate subjectivity. The viewpoints of 100 experts were gathered in matrix form, ensuring that the consistency of their opinions was maintained at less than 10%. Subsequently, FAHP was employed to calculate priority weights among the parameters, forming the basis for the development of rating equations. The developed SHM tool was applied to 12 URM structures including historical buildings, residential buildings, and office buildings in different locations.

Bidirectional purification of nonlinear guided waves using metamaterial filters for structural health monitoring

LLM-based multi-agent system for dam structural health monitoring

DamCrack: A drone and smartphone image dataset for 2D and 3D damage assessment in concrete dams.

Accurate detection of dam cracks from Unmanned Aerial Vehicle (UAV) imagery is crucial for structural health monitoring. However, prevailing methods face significant challenges in achieving a balance between the precise identification of minuscule cracks and computational efficiency when processing high-resolution images. To overcome these limitations, this paper proposes HiResDC-YOLO, a novel deep learning framework based on an enhanced YOLOv12 architecture. Our main contributions are threefold: First, we introduce an Adaptive Dynamic Transformer (ADyT) module to strengthen nonlinear feature representation and stabilize gradient flow during training. Then, we design a Multi-Scale Enhanced Detection (MSED) head that effectively utilizes shallow, high-resolution features to significantly improve the detection capability for fine cracks. Besides, we develop a Multi-Scale Convolutional Attention (MSCA) module to capture comprehensive contextual information across different scales by integrating deep convolutional layers and a multi-branch fusion mechanism. Furthermore, we propose a High-Resolution Adaptive Inference (HiResInfer) strategy, which utilizes region-guided slicing and feature caching to dramatically accelerate the inference speed on full-resolution images without compromising the integrity of small targets. Extensive experiments on a challenging self-collected UAV dam crack dataset demonstrate that HiResDC-YOLO achieves state-of-the-art performance, surpassing existing methods significantly in terms of precision, recall, and mean Average Precision (mAP), while maintaining high computational efficiency. This work presents a robust and practical solution for real-time dam inspection and engineering safety monitoring. The code and related resources are available at: https://github.com/lijin6/HiResInfer-YOLO.git.

ABSTRACT Cable force estimation is critical for ensuring the structural health of cable-stayed bridges. This study assesses the accuracy of three vibration-based models – classical string theory, a least-squares beam formulation, and a two-mode beam combination – using lift-off test results as ground-truth validation. Field measurements were analysed to extract natural frequencies, which served as inputs to predict axial forces. The correlation between estimated forces and lift-off measurements was evaluated using the Pearson coefficient and mean absolute error (MAE). Results show strong agreement between vibration-based estimates and lift-off forces, with correlation coefficients ranging from 0.8610 to 0.8619. Beam-theory approaches outperformed the string model, achieving 7–8% lower MAE confirming that incorporating bending stiffness and multi-mode effects improves accuracy. However, small numerical differences suggest axial tension remains the dominant factor, with flexural rigidity having minor influence on long and slender cables. The near-perfect correlation between the beam-based models (r > 0.9996) highlights their internal consistency. Overall, these findings demonstrate that vibration-based models are reliable tools providing essential insights for structural health monitoring and maintenance strategies.

Artifact suppression in ultrasonic guided wave damage imaging enhanced by topologically optimized mode selective meta-filter.

ABSTRACT Polyethylene (PE) pipelines are critical for gas transportation, but long-term service leads to wall deterioration, leading to stress concentration and subsequent ductile failure. Ultrasonic testing is effective formaterial and structural damage monitoring. However, current approaches exhibit limited capability for monitoring damage evolution in PE pipes. This study proposes an ultrasonic-based structural health monitoring framework to characterise the ductile failure of PE pipes under high hoop stresses. Online ultrasonic measurement was implemented during accelerated ageing tests using water-coupled transducers to capture real-time degradation processes. Two damage indices – the Energy-Based Damage Index (EDI) and Nonlinear Signal Difference Coefficient (NSDC) – were extracted from ultrasonic echo signals on the back wall of the pipes. The Criteria Importance Through Intercriteria Correlation (CRITIC) weighting method was employed to integrate them into a Comprehensive Damage Index (CDI). Experimental results demonstrate that both EDI and NSDC exhibit sensitivity to ductile damage evolution in PE pipes, effectively monitoring wall damage. The CDI achieves more comprehensive damage characterisation by accounting for ultrasonic energy attenuation and waveform distortion induced by material deterioration. The proposed framework provides a validated methodology for monitoring of ductile damage in PE pipes, bridging the gap between static defect identification and dynamic condition assessment.

ABSTRACT This study presents the first comprehensive evaluation of SouthPAN-SBAS-aided real-time precise point positioning (RT-PPP) for high-rate applications. A single-axis shake table was used to simulate harmonic oscillations and earthquake-induced ground motions, and GNSS data was collected at 20 Hz. Solution accuracy was evaluated by comparison with relative GNSS positioning, post-processed PPP, conventional RT-PPP, and Linear Variable Differential Transformer (LVDT) reference. Results indicate that SBAS-aided RT-PPP accurately captures oscillation frequencies and dynamic displacements. These findings demonstrate the strong potential of SouthPAN SBAS–assisted RT-PPP for GNSS seismology and structural health monitoring in the Australia–New Zealand and broader Asia–Pacific region.

Elastodynamics-encoded recurrent neural networks for characterization of anisotropic elastic constants and stiffness degradation.

Abstract Flexible piezoresistive sensors are attractive for structural health monitoring; however, the simultaneous optimisation of sensitivity, mechanical compliance, and response speed remains challenging. In this study, room-temperature vulcanized methyl silicone rubber (RTV)/carbon black (CB)/multiwalled carbon nanotube composite sensors are fabricated by solution blending, and the coupled effects of a silane coupling agent (KH-550) and RTV-matrix viscosity on the mechanical, electrical, and sensing performances of the composite are systematically clarified. KH-550 reduces the percolation threshold, lowers bulk resistivity by up to 26.54% at 15 phr CB, and enhances pressure sensitivity by 48.45%, while inevitably increasing hardness. Matrix viscosity is shown to be a key regulator of the dynamic response: decreasing viscosity from 1500 to 300 mPa·s shortens both response and relaxation times by more than 50%. By combining experiments and ABAQUS simulations, we demonstrate that the Poisson effect is associated with anomalous piezoresistive behaviour when the conductive fillers are near the percolation threshold and are poorly dispersed. Finally, we introduce a dimensionless figure-of-merit F that integrates normalised sensitivity, hardness, and response speed, and construct a bubble chart that identifies a practical window (CB ≥15 phr, moderate viscosity and 3 phr KH-550) for balanced performance. This study provides a reference for the design and compositional optimisation of functional materials for high-performance flexible piezoresistive sensors.

Bridge infrastructure worldwide is aging, while demands for safety, sustainability, and cost-efficiency intensify challenges in bridge management (BM). This study develops a comprehensive knowledge map of BM through scientometric analysis, clarifying its intellectual structure, technological frontiers, and emerging themes. Using CiteSpace, 1582 peer-reviewed articles (2003-2024) from the Web of Science Core Collection were analyzed, with keywords and gaps refined via a Delphi process involving five senior experts. Scientometric techniques-including co-citation, cluster mapping, citation bursts, and keyword time-zone analysis-were employed to trace field evolution. Results reveal three stages: exploratory (2003-2009), steady growth (2010-2016), and digital transformation (2017-2024). Ten thematic clusters emerged, spanning intelligent management, hybrid decision-making, inspection technologies, and life-cycle sustainability. Research hotspots highlight digital twins, machine learning, condition assessment, and structural health monitoring. Five evidence-informed future priorities were identified: integrated AI-driven frameworks, hybrid decision models, network-level inspection planning, sustainability-oriented life-cycle assessment, and robust resilience modeling. Co-authorship and institutional collaboration networks further reveal global hubs shaping BM knowledge exchange. This study provides an updated mapping of BM, offering a structured agenda that links quantitative evidence with practical directions for advancing intelligent, sustainable, and risk-resilient BM.