Project: #10 Infrastructure Monitoring for Gradual Damage Detection Form an Inservice Light Rail Vehicle Progress Report - Reporting Period Ending: Sept. 30, 2017 Principal Investigator: Jacobo Bielak Status: Active Start Date: Jan. 1, 2017 End Date: Aug. 31, 2018 Research Type: Applied Grant Type: Research Grant Program: UTC FAST Act Grant Cycle: 2017 TSET UTC Progress Report (Last Updated: Oct. 5, 2017, 10:02 a.m.) % Project Completed to Date: 25 % Grant Award Expended: 0 % Match Expended & Document: 0 USDOT Requirements Anticipated Research Outcomes We have developed a statistical algorithm for indirect structural health monitoring of railway bridges and other similar rail transit structures that is capable of detecting a gradual change of structural integrity. Our approach is able to extract reliable indicators of damage from acceleration signals recorded onboard a vehicle moving over the bridge. Beyond introducing this novel and breakthrough approach, we expect to contribute to the scientific literature in this domain by providing a physics-based explanation for the effectiveness of our statistical method, as well as by providing simulation and experimental observations to corroborate these findings and validate the performance of the algorithm. For this first outcome, we expect to submit and publish our findings in a top academic journal (e.g., Mechanical Systems and Signal Processing), as well as through oral presentations at conferences and other professional gatherings. We will then focus our attention towards larger-scale structures and testbeds including vehicular and train bridges in the city of Pittsburgh, to refine and test the applicability of our approach in realistic settings. Anticipated Impacts The key distinguishing aspect of our approach in contrast with other indirect monitoring approaches for bridges is that rather than pre-define the features we extract from the sensor acceleration signals, the features we use are automatically defined by the algorithm during the training process, and thus does make assumptions about the generalized applicability of any pre-defined feature to different structural/environmental settings. We anticipate that our approach will open a new research direction for advancing the practice of indirect structural health monitoring, and could finally allow for the implementation and use of these approaches in practice. Did research results confirm or change practice? Our results have the potential of radically changing the way railway bridges and other related rail transportation structures are inspected and monitored. Web Links N/A Issues The remaining challenges to be addressed in the following project period include 1. expanding and validating our approach to be robust to various levels and types of noise. For this purpose, we plan to utilize both numerical simulations and laboratory experiments 2. testing the sensitivity of our approach to the intensity and type of damage. 3. large-scale evaluation with the field data collected from the Pittsburgh light rail system that we already installed through previous years' funding. Accomplishments 1. Demonstrated state-of-the-art breakthrough results on unsupervised gradual damage detection and estimation algorithm for vehicle-based bridge and rail track damage diagnosis 2. Presented our work at PIANC this Fall 3. Engaged in discussions with Arup and Union Pacific for follow-on funding opportunities. In discussion to submit a proposal to other agencies such as FRA and TRB 3. Drafted a journal paper on unsupervised damage diagnosis, to be submitted Fall 2017 4. Drafted a journal paper on the light rail train response data collected from the indirect monitoring system we developed through the collaboration with the Port Authority of Allegheny County 5. The project has been introduced and utilized for a graduate level project course on sensing and data mining for smart structures and systems. This project provided hands-on experience to the students and helped with engaging/inspiring them in transportation safety-related projects. 6. The work is presented through multiple invited seminars at Stanford, CalTech, Princeton, Georgia Tech, etc. to promote collaboration and expansion of the work. 7. Invited as a panel at the Smart Cities Summit to introduce this project and potential applications to and collaboration with the USPS.