Date and time: October 15th, 2020 at 15h30
Track: Structural Engineering
Topic: Vibration-based damage detection of the Haringvlietbrug: A data-driven approach.
Description: Bridges can be considered as the most critical veins in our built environment, especially in our 24-hour economy. The construction of many bridges in The Netherlands, date back to 75 years ago and have been in use ever since. Rijkswaterstaat, the administrator of the Dutch infrastructures, faces the great task of keeping the infrastructures safe and structurally sound. This includes the monitoring of over a 1000 bridges. Instead of performing visual inspections on an incidental basis, possibly detecting damage months after development, a transition is made towards online monitoring of bridges with the use of sensors attached to the bridge. These sensors continuously measure the bridge vibrations generating GBs of data every day.
During this thesis research, a total of 32 sensors have been installed on the Haringvlietbrug with the aim to detect small-scale fatigue cracks between the transverse beams and longitudinal stiffeners (bulb profiles). Multiple analyses have been conducted including advanced preprocessing with clustering, feature extraction and dimensionality reduction with principal component analysis and relationship learning with artificial neural networks. To find out if damage indeed was detected and what challenges we faced during the research, please join the defence online on October 15th at 15:30!