Structural Health Monitoring and Prognostic of Industrial Plants and Civil Structures: A Sensor to Cloud Architecture

The deployment of Structural Health Monitoring (SHM) systems is a natively interdisciplinary task that involves joint research contributions from sensing technologies, data science and civil engineering. The capability to assess, also from remote stations, the working conditions of industrial plants or the structural integrity of civil buildings is widely requested in many application fields. The technological development aims to continuously provide innovative tools and approaches to satisfy these demands. As a first instance, reliable monitoring strategies are needed to detect structural damages while filtering out environmental noise. Ongoing solutions to tackle these topics are based on the exploitation of highly customized sensing technologies, such as shaped transducers for Acoustic Emission (AE) testing or Micro-Electro-Mechanical System (MEMS) accelerometers for Operational Modal Analysis (OMA) [1]. On the other hand, effective data acquisition and storage techniques must be employed to cope with the heterogeneity of the sensing devices and with the amount of data produced by collecting raw measured signals. Finally, damage detection and prediction tasks should be computed via data-driven algorithms that can complement the model-based alternatives traditionally used in civil engineering. Layered SHM architectures [2] represent straightforward approaches to address the system complexity originated by this interdisciplinary design; however, few real-world implementations have been presented so far in the literature. In this paper, we overcome these limitations by presenting an Internet of Things (IoT)-based SHM architecture for the predictive maintenance of industrial sites and civil engineering structures and infrastructures. The proposed cyber-physical system includes a monitoring layer, that consists of accelerometer-based sensor networks, a data acquisition layer, built on the recent W3C Web of Things standard [3], and a data storage and analytics layer, which leverages distributed database and Machine Learning tools. We extensively discuss the hardware/software components of the proposed SHM architecture, by stressing its advantages in terms of device versatility, data scalability and interoperability support. Finally, the effectiveness of the system is validated on a real-world use-case, i.e., the monitoring of a metallic frame structure located at the SHM research labs of the University of Bologna, Italy, within the MAC4PRO project [4].