Machine Learning for Structural Health Monitoring: Closing the Gap between Research and Industrial Applications

Dr. Fuh-Gwo Yuan
Department of Mechanical and Aerospace Engineering
North Carolina State University, Raleigh, NC

Abstract:

A physics-based approach to structural health monitoring (SHM) has practical shortcomings which restrict its suitability to simple structures under well controlled environments. With the advances in information and sensing technology (sensors and sensor networks), it has become feasible to monitor large/diverse number of parameters in complex real-world structures either continuously or intermittently by employing large in-situ (wireless) sensor networks. The availability of this historical data has engendered a lot of interest in a data-driven approach as a natural and more viable option for realizing the goal of SHM in such structures. However, the lack of sensor data corresponding to different damage scenarios continues to remain a challenge. Most of the supervised machine-learning/deep-learning techniques, when trained using this inherently limited data, lack robustness and generalizability.

Physics-informed learning, which involves the integration of domain knowledge into the learning process, is presented here as a potential remedy to this challenge. The concept of physics-informed neural networks (PINNs) will be presented using two kinds of problems: (1) Forward problem: scattered wavefield reconstruction in complex aerospace structures from sparse sensor data. It is shown that honoring the underlying physics and/or domain knowledge during the training process of ANNs leads to improved robustness and better generalization. By doing so, diffraction limit can be transcended thereby achieving super resolution imaging. (2) Inverse problem: detailed damage characterization of complex composite structures using deep learning. Lastly, a recent vision-based SHM system using the digital image correlation (DIC) to capture scattered ultrasonic wavefield for image the damage was developed. Leveraging the physics-informed learning with the vision-based SHM is in progress toward damage characterization.

Bio:

Dr. Yuan has been with North Carolina State University since 1989. Currently he is a full professor at Mechanical and Aerospace Engineering, NC State. Currently he serves as a Langley Professor at National Institute of Aerospace. His recent research includes structural health monitoring/management (SHM), nondestructive inspection (NDI), machine learning, multi-functional materials, nano/meso scale sensors, advanced computing tools with smart sensors, damage prognosis, and energy harvesting.