New Sensor Nodes, Cloud, and Data Analytics: Case Studies on Large Scale SHM Systems

The progress in the world of new sensors is running fast, offering good performances, and reliable solutions with initial costs orders of magnitude lower than those faced only a few years ago. The spread of new electronic devices, like microcontrollers, the increasing power of the networks for data transmission and management and in the end the availability of the new data-driven approaches have created a revolution in SHM approaches, not yet fully mastered. The way to design a SHM system is going to be deeply revised in an industrial perspective, within a complex framework in which everything has to be planned into details since the beginning, including the development of a metrological culture, the personnel education, the need of spare parts, re-calibration, …. This also means a revolution in data management: huge data flows not only create hardware problems related to their transfer; the software too requires a great deal of effort to compress data, also due to the actual cost of cloud resources. All these facts, accounting for the real metrological performances of the best MEMS sensors available at present, also require simplified data analyses, as software complexity is now mainly transferred to the network management. A trade-off must be looked for between the big redundancy offered by the actual networks and the need of a simple and prompt information, granting the structure safety: That is why as the data rates increase, the algorithms to be adopted must be simple, reliable, eventually adapted to edge computing at the sensor level, where hardware power is now present though at a reduced scale. The chapter shows such an approach in a real case from the system design, its birth, and its proper use for damage detection, up to the detection of a structural failure.