Conference Agenda

Continuous bridge monitoring often relies on fixed-frequency data acquisition, which leads to redundant measurements and unnecessary power consumption under varying environmental conditions. This study proposes a hierarchical adaptive sampling strategy for the health monitoring of bridges, developed to balance energy efficiency and response sensitivity under changing environmental and traffic conditions. At the sensor-node level, each lightweight monitoring unit combines a Raspberry Pi with a vision-based strain sensor and a temperature–humidity module. The node automatically adjusts its sampling rate in response to real-time weather information, as retrieved from meteorological data delivered by an online service, and measured strain variations. When the wind level reaches a critical (strong-wind) threshold, the node switches to a high-frequency mode to capture the induced vibrations; otherwise, the node reduces its sampling rate to save energy. Furthermore, moderate or adverse weather such as rain, snow, or dust storms prompts a temporary rate increase to ensure data reliability. This approach allows the monitoring frequency to vary with the environment, providing efficient performance at minimal power cost. At the system coordination level, a cloud-based control layer built on the MQTT protocol manages real-time communication among sensor nodes. The primary node analyses strain signals using a sliding-window method to assess the rate of change and fluctuation intensity. When several consecutive windows exceed predefined thresholds, indicating either strain exceedance or frequent vehicle-induced vibrations, it then sends an alert to the cloud. The cloud then instructs nearby nodes to raise their sampling frequency for synchronised, high-resolution monitoring. Once the strain variations and event rate stabilise below threshold values, the system reverts to its low-frequency mode, maintaining long-term energy balance. The proposed framework has been implemented and tested on an in-service bridge. Field results show that it accurately captures vehicle- and wind-induced strain responses, while reducing the total energy consumption by about 83% compared with fixed-frequency acquisition. The study demonstrates that adaptive sampling effectively enhances the intelligence, autonomy, and energy efficiency of bridge monitoring systems, contributing to the development of smarter and more sustainable monitoring solutions.

Ensuring the integrity of critical infrastructure, such as bridges, dams, and large-scale structures, is essential to safety, reliability, and operational continuity. These assets are exposed to mechanical, thermal, environmental, and operational loads that can accelerate fatigue, corrosion, wear, and other deterioration mechanisms. Traditional inspections are periodic and costly, while sensor-based monitoring systems generate large volumes of imperfect field data that are difficult to interpret without engineering context.

This paper presents a physics-informed machine learning and multi-sensor fusion framework for fatigue-oriented Structural Health Monitoring (SHM) and predictive maintenance. The method is demonstrated on an instrumented steel-concrete composite bridge using resistive strain, fiber Bragg grating (FBG) strain and temperature, and displacement measurements. In decision-critical SHM applications, uncertainty arises from noisy sensor signals, thermal effects, drift, missing data, and ambiguous operating events. The proposed framework reduces this uncertainty by applying physics and engineering principles to structure the correction, validation, and interpretation of sensor data.

Machine learning supports thermal compensation, anomaly screening, cross-sensor substitution, and confidence tagging, while deterministic engineering methods remain responsible for strain interpretation, event extraction, rainflow-style cycle counting, and fatigue damage indicators. Results show that multi-year monitoring data can be reduced into compact fatigue-relevant features while preserving traceability to raw measurements. A supervisory agentic layer coordinates data-quality checks, multi-sensor consistency review, and confidence-tagged substitution, creating an auditable workflow for engineering decision support. SHM value does not come only from better sensors or better ML models. It comes from traceable workflows that turn imperfect field data into reliable, reviewable, and actionable engineering evidence.

In the field of structural health monitoring (SHM) of steel structures, early detection of fatigue damage in welded joints is still considered a crucial issue and represents a major challenge. This study explores the use of data fusion from multiple sensor types to improve the early identification of fatigue-related damage in welded assemblies. Standardized fatigue tests were conducted on steel specimens equipped with diverse sensors. Crack detection was performed consistently across all sensors using a unified methodological framework, with the Mahalanobis distance applied to assess detection performance. Individual sensor data were analyzed to detect and localize cracks, followed by a feature-level data fusion approach employing mathematical and statistical tools, including the Mahalanobis distance, to enhance detection earliness. Two fusion strategies were evaluated: intra-sensor fusion (within the same sensor type) and inter-sensor fusion (across different sensor types). Results indicate that intra-sensor fusion significantly enhances early detection, whereas inter-sensor fusion provides limited improvements, highlighting the need for further optimization.