With the rapid growth of wireless IoT in structural health monitoring (SHM), distributed sensing systems using edge computing have become increasingly popular for their scalability, easy deployment, and low cost. However, since each sensor node has its own clock, this decentralized setup leads to time drift and synchronization errors. Asynchronous signals cause phase misalignments, reducing the accuracy of vibration-based modal analysis and the reliability of damage detection. Therefore, accurate SHM requires synchronized measurements across all sensor nodes. Conventional clock protocols such as the Timing-Sync Protocol for Sensor Networks (TPSN), Reference Broadcast Synchronization (RBS), and Flooding Time Synchronization Protocol (FTSP). These approaches synchronize the local clocks of distributed sensor nodes, ensuring consistent timing across the network. Nevertheless, despite achieving clock alignment, they cannot guarantee that data acquisition occurs simultaneously across all nodes. Variations in communication latency, processing delays, operating system scheduling, and sensor readout times can still cause misalignment among recorded samples or events. This is particularly critical in vibration-based SHM, where acceleration responses are captured at high sampling rates over long durations. Even sub-millisecond offsets can introduce phase errors, potentially leading to inaccurate damage evaluation. Clock-layer synchronization protocols are often inadequate for high-precision applications, making post-processing synchronization essential for accurate temporal alignment in distributed sensing systems. Various techniques have been developed to estimate time lags between asynchronously acquired signals. Lei et al. used an Auto-Regressive Moving Average Vector (ARMAV) model, while Wang et al. applied a Stochastic Subspace Identification (SSI) approach based on state-space modeling. Zhou et al. proposed a frequency-domain method that converts time lags into phase shifts at modal frequencies, and Dragos et al. estimated lags using cross-spectral density phase slopes. Fu et al. addressed multi-mode phase variations by mapping time lags to phase-period differences within a defined bandwidth, using Variational Mode Extraction (VME) to isolate fundamental modes. However, traditional time-domain methods depend heavily on model order selection, which becomes difficult for large or complex systems and can reduce estimation accuracy. Frequency-domain methods alleviate this to some extent but remain sensitive to noise, limited by bandwidth, and affected by modal coupling and nonlinear phase behavior. They also rely on simplifying assumptions, such as zero phase for the fundamental mode and accurate identification of modal frequencies. Consequently, their performance often degrades under sub-sample timing errors common in real-world SHM applications. To overcome these limitations, this study introduces a Time-Lag Convolutional Neural Network with Attention (TLCA), the first deep learning-based method for time lag estimation and data synchronization in wireless SHM systems. The proposed end-to-end framework learns temporal dependencies and cross-sensor correlations directly from raw time-series data, enabling accurate lag estimation without prior assumptions or mathematical simplifications. A resampling module with Bayesian optimization further refines alignment for non-integer time lags, achieving sub-sample synchronization. The model was validated on both simulated and real-world datasets, demonstrating robust performance under varying noise levels, nonlinear dynamics, and a wide range of time lags.

Tower-radar computer vision (TRCV) represents an emerging application-oriented field of study. Here, image-type measurements are acquired from radar transceivers bound to the mast of wind power turbines and these radargrams subsequently get analyzed using data-driven algorithms. TRCV shows promise for reducing downtime and increasing safety of such renewable energy plants; accordingly, the respective monitoring systems should long-term perform real-time recognition of moving objects like rotor blades and their condition, disregard maintenance workers, identify birds and bats or unauthorized aerial vehicles, in order to trigger appropriate multi-faceted responses. This article focuses on the radar-only computer-vision task of classifying rotors without complementary costly instrumentation [1]. Progress in TRCV for blade monitoring remains, however, hindered by the limited publicly available data [2]. For this specialized task, recently [3] general-purpose feature extractors have proven robust to environmental and operational conditions and as a valuable building block in TRCV processing pipelines. Such large models, like OpenCLIP, have been pre-trained on internet-scale amounts of text-image pairs. They however (i) commonly still require additional data for transfer to dedicated use cases, as well as (ii) exhibit power demands or latencies incompatible with real-time resource-constrained application scenarios; models hence need to be compressed. In this paper, measured radargrams from field experiments are therefore complemented with the novel synthetic dataset SiWiRoRa as well as further open imagery. Here, several specialized image-type datasets are carefully compiled to span a context around the focal measured radargrams. Second, the main geometrical directions of this context landscape are explained by classical image statistics and with human perceptions collected from annotators. Third, large models are distilled towards lightweight capacity where it is found that both the synthetic dataset but also seemingly unrelated imagery can improve performance in blade classification. Accordingly, routes for enhancing the SiWiRoRa dataset are suggested and moreover implemented so as to further advance TRCV.

References

[1] Alipek, Sercan, et al. "Potential and Limitations of Anomaly Detection via Tower-Radar Monitoring of Wind Turbine Blades in Regular Operation with Convolutional Networks." EWSHM (2024).

[2] Mälzer, Moritz et al. "Radar-based structural monitoring of wind turbines blades: Field results from two operational wind turbines." IWSHM (2023).

[3] Kexel, Christian et al. "Mast-Bound and Too Curious: Overcoming Drift in Wind-Tower Radar for Blade Monitoring Using Pre-Training, Augmentation and Weight Consolidation Due to Correlated Conditions." LATAM-SHM (2026)

Pavement condition forecasting remains limited by the reliance on costly structural measurements and curated tabular inputs, hindering scalability in large road networks. We investigate whether accurate multi-year forecasting can instead be achieved directly from raw surface geometry acquired in routine surveys. We present the first end-to-end pipeline forecasting annual Pavement Condition Index evolution from Laser Crack Measurement Systems (LCMS) data. Experiments on French national road network surveys demonstrate substantial improvements in spatial alignment and stable multi-year PCI forecasting using surface geometry alone, establishing LCMS data as a viable foundation for scalable pavement deterioration modeling.

Fatigue evaluation of steel railway bridges often focuses on cross-girder to main beam connections, where high stress concentrations can lead to progressive damage. Conventional S–N curve–based fatigue assessment methods typically rely on simplified loading assumptions and deterministic damage accumulation, which may not adequately capture the complex multiaxial stress states and evolving traffic conditions experienced by in-service railway bridges. This study presents a probabilistic framework for fatigue life estimation that explicitly incorporates realistic traffic variability using bridge weigh-in-motion data. The proposed approach integrates multiaxial fatigue criteria, rainflow cycle counting, and Miner’s rule with a data-driven surrogate model based on Gaussian Process Regression (GPR) and Monte Carlo simulation. Unlike traditional approaches that assume fixed loading spectra, the GPR model is trained on incremental fatigue damage progression, enabling probabilistic prediction of Remaining Useful Life (RUL) under future traffic growth scenarios. A UK steel railway bridge is investigated using a detailed OpenSees finite element model combined with measured traffic data. The results demonstrate that the framework captures fatigue progression trends, quantifies uncertainty in RUL estimation, and identifies critical stress locations under realistic operational conditions. By integrating physics-based fatigue modelling with data-driven probabilistic prediction, the proposed method provides a practical tool for traffic-informed life assessment, supporting risk-based structural health monitoring and long-term asset management of railway bridges. Figure 1 illustrates the workflow of the proposed framework.

The assessment of road conditions is essential for maintaining the safety, functionality, and longevity of transportation infrastructure. However, traditional evaluation methods often suffer from subjectivity, high operational costs, and delayed response times. To overcome these limitations, this study presents an autonomous, AI-enabled system for comprehensive pavement condition monitoring using crowdsourced RGB-D data. The proposed vehicle-mounted data acquisition platform enables affordable collection of 3D pavement surface information during regular driving operations. By leveraging RGB-D sensors, the system captures both 2D color and 3D depth data across full lane widths, with accurate spatial alignment ensured by high-precision GPS integration. Pavement conditions are evaluated using the Pavement Surface Evaluation and Rating (PASER) system for asphalt pavements as a case study. Deep learning and computer vision models classify pavement surfaces into eight categories, including healthy surface, open joint, manhole, crack sealant, transverse crack, longitudinal crack, alligator cracking, and pothole. Quantitative analysis further supports detailed distress characterization, defect tracking, and maintenance evaluation, offering a scalable and data-driven framework for real-time pavement management and rehabilitation quality control. More than 2,500 miles of roads are evaluated using this system.