Structural Health Monitoring (SHM) is an essential process that involves real- time monitoring of the physical condition of a mechanical structure in the presence of environmental variations. This monitoring relies on data collection through sensors and the utilization of reference models that describe the structure in its initial state. This coupling between sensors and numerical models proves to be extremely challenging, primarily due to the significant disparity between the limited number of available sensors and the high complexity and dimensionality of the models required for accurate monitoring. The fondamental objective of this research project is to improve state-of-the-art SHM strategies coupling experimental data with numerical modelling by combining them with physics-informed neural networks (PINNs). The numerical model should assist the PINN in enhancing limited real-world data with model-generated, damage-sensitive physical features. This integration aims at generalizing SHM methods while making them adaptable to various dynamic conditions and improve their robustness to noise, sensor defects, and model errors.
Persons Involved:
- Laurent Mevel, Inria, PI
- Christophe Droz, Inria, ISFP, Co-PI
- Adrien Mélot, SRP Inria
- Alvaro Gavilán Rojas, Phd Student, Inria
- Subhamoy Sen, IIT Mandi, PI
- Nikkil Mahar, Phd Student
- Shereena OA, Postdoc
Meetings:
Kick off at IIT Mandi in September 2024
