SHM
Anomaly detection in structural health monitoring - a collaboration with Isterre UGA
This study embodies research efforts in automatically detecting out-of-distribution data outlier in real-time for autonomous decision making for complex systems.
This an ungoing collaboration work with Isterre UGA, to propose an environment-informed approach to probabilistically detect the potential structural damage. Several environmental factors, shown below and three vibrational modes are considered in this anlysis. Each mode is associated with its natural frequancy $\omega$ and damping ratio.
A glimpse of Environmental factors.
We will detect the potential structural damage based on the 95% prediction interval. Three examples with respect to three modes are shown below:
Three examples with respect to three modes