Location:
on-site: complex room 1006
digital: MS Teams (link shared via e-mail)
Fabian Dethof: Simulating “semi-guided” elastic wave propagation in concrete – understanding Impact Echo data
The usage of low-frequency signals and easy data acquisition make Impact Echo a widely applied NDT method in the field of civil engineering ever since its introduction in the 80’s. However, the physical principle of the method was not fully understood until 2005. Numerical simulations are used to better understand acquired datasets and improve data evaluation by introducing new evaluation procedures and using existing machine learning methods.
Lizzie Neumann: Confounder-adjusted Covariances of System Outputs and Applications to Structural Health Monitoring
Automated damage detection is integral to structural health monitoring (SHM) systems. However, changes in the data result not only from damage but also from environmental or operational influences. Consequently, it is necessary to determine the confounding factors and remove their effects from the measurements or extracted features. Methods used so far, however, neglect potential changes in higher-order statistical moments, although the output covariances are essential for generating reliable diagnostics for damage detection. We propose an approach that explicitly quantifies changes in the covariance using conditional covariance matrices, and we apply the method to
real-world bridge data. Our results show that temperature changes affect the covariances of sensor measurements and natural frequencies. By combining our new approach with standard methods
for damage detection, we can generate more reliable diagnostic values and fewer false alarms.