An IoT-AI Toolchain for Structural Health Monitoring

Next-generation Structural Health Monitoring (SHM) systems can benefit significantly from the integration of Internet of Things (IoT) and Machine Learning (ML) technologies to assess the integrity of civil buildings and infrastructures. However, the deployment of scalable and accurate data analytics techniques for damage detection and classification is often hindered by the processing of voluminous sensor data and by the complexity of generating representative pseudo datasets for actual healthy and damaged conditions. We propose a framework for SHM data management and analytics that draws upon competencies and tools from various domains, including IoT, numerical modeling, and AI. Our methodology spans from data gathering from on-field sensors, data processing upon their physical interpretation, up to an alarming system. It integrates data-driven and model-based approaches to perform anomaly detection and, consequentially, to classify damage characteristics via a set of dedicated algorithms. In this work, we specifically describe the IoT-AI architecture for SHM, highlighting critical points and further developments. In addition, we present an illustrative implementation carried out on a concrete structure by reporting examples of possible outcomes of our framework.