PI: Mario Bergés
Co-PI(s): Katherine Flanigan
University: Carnegie Mellon University
Industry partner: Wabtec
Using acceleration data from in-service trains has become a popular track inspection approach because it is a low-cost way to monitor rail tracks more frequently. However, the measurements collected in this manner demand a more complex analysis process given that they only indirectly reflect the phenomena of interest and are heavily influenced by the vehicle-track interaction. Our team has had a successful long-term collaboration with the Port Authority of Allegheny County to develop a series of algorithmic innovations resulting in a learning-based strategy that can adaptively find ways to detect, localize, and quantify sudden or gradual damage on tracks, validating it in simulation, lab-scale experiments, and on real passenger trains. Here we propose to expand the approach into a more holistic model that can dynamically decouple the stationary-mobile agent interaction to separately characterize the condition of the stationary asset (tracks) and the condition of the mobile asset (train) via the addition of a new piece of long-range vibrometry equipment that will enable the team to effectively measure acceleration across large expanses of track at a reduced cost. A new partnership with Wabtec Corporation will also allow us access to new/different testbeds and expertise.