A crowdsensing platform for structural health monitoring of rural bridge infrastructure

Ageing bridge infrastructure in rural areas poses significant safety challenges, yet traditional structural health monitoring systems remain prohibitively expensive for regions with limited budgets. The Horizon Europe FUTURAL project developed an innovative crowdsensing platform to address this gap, enabling affordable infrastructure monitoring through vehicle-based indirect measurements combined with low-cost Internet of Things sensors.

The Resilience to Shocks Smart Solution integrates multiple components: a mobile application (FUTURAPP) connecting USB-C accelerometers to smartphones for data acquisition; cloud-based processing infrastructure for secure data transmission and storage; an unsupervised autoencoder neural network trained on frequency-domain acceleration data (2–15 Hz) for the detection of deviations from a learned baseline structural response; finite element models developed using ANSYS APDL for structural validation; and a web-based dashboard featuring Structural Health Index visualizations and citizen engagement tools. The platform was deployed and tested on two pilot bridges: a three-span concrete road bridge in Durangaldea, Spain, and the Diakofti Bridge in Kythira, Greece.

The autoencoder algorithm successfully distinguished between measurements consistent with the learned baseline acceleration response and signals corresponding to operating conditions clearly outside the monitored bridge response, using reconstruction error as the anomaly-sensitive feature, with the 99th percentile threshold effectively discriminating measurements consistent with normal operating conditions from those clearly outside it. Finite element models showed good agreement with in-situ vibration frequency measurements. The web platform demonstrated high usability among non-technical users, with the intuitive Structural Health Index gauge enabling immediate comprehension of bridge conditions. Citizen issue tracking features facilitated community participation in infrastructure safety monitoring.

The crowdsensing methodology offers significant advantages over traditional approaches in terms of cost, scalability, and accessibility for resource-constrained regions. The results presented should be interpreted as demonstrating the feasibility of an anomaly detection framework within a deployable monitoring platform rather than as a fully validated damage-detection methodology, since no measurements corresponding to real damage scenarios are currently available for the monitored bridges in this study. The platform represents a meaningful advancement toward digitalisation of rural infrastructure management, with the methodology and tools providing a foundation for progressive adaptation, broader validation campaigns, and methodological refinement before larger-scale deployment across European transport networks.

Bridges in rural areas are essential for daily life as they connect villages, allow farmers to transport goods, and enable emergency services to reach people in need. However, many of these bridges are aging and need regular check-ups to ensure they remain safe. Traditional monitoring systems are expensive and require installing many sensors directly on each bridge, which is not practical for rural regions with limited budgets and hundreds of small bridges.

In this project we developed a more affordable solution. Instead of placing sensors on every bridge, we put a small, low-cost accelerometer (a device that measures vibrations) inside a vehicle connected to a smartphone. When the vehicle drives over a bridge, the sensor records how the bridge vibrates. These vibration patterns can reveal a healthy bridge that has a normal pattern, while a damaged bridge will show irregularities.

The collected data is sent to a cloud infrastructure or data lake, where an artificial intelligence system learns what normal bridge vibrations look like. When new measurements arrive, the AI compares them to what it learned and flags any unusual patterns that might indicate damage. Asset managers and citizens can then view results through a user-friendly web platform that shows in real-time the health status of the bridges. This system is tested on two real bridges: one in the Basque Country of Spain and another on the Greek island of Kythira. Both bridges are critical for their local communities and face risks from earthquakes, heavy rainfall, and floods. Our tests showed that the system can successfully distinguish between normal conditions and potential problems, offering a practical way for rural communities to monitor their bridges affordably, helping keep roads safe without requiring expensive equipment on every structure.