A0539
Title: Monitoring of confounder-adjusted scores using conditional principal component analysis
Authors: Philipp Wittenberg - Helmut Schmidt University (Germany)
Martin Koehncke - Helmut Schmidt University (Germany)
Alexander Mendler - Technical University of Munich (Germany)
Sylvia Kessler - Helmut Schmidt University (Germany)
Jan Gertheiss - Helmut Schmidt University (Germany)
Lizzie Neumann - Helmut Schmidt University (Germany) [presenting]
Abstract: In structural health monitoring (SHM), measurements from various sensors are collected and reduced to damage-sensitive features. Diagnostic values for damage detection are then obtained through statistical analysis of these features. The system outputs, i.e., sensor measurements and/or extracted features, however, depend not only on damage but also on confounding factors (environmental or operational variables). These factors affect not only the mean but also the covariance. This is particularly significant because the covariance is often used as an essential building block in damage detection tools. A method is presented for calculating confounder-adjusted scores utilizing conditional principal component analysis, which entails estimating a confounder-adjusted covariance matrix. The technique is applied to monitor real-world data from the Vahrendorfer Stadtweg bridge in Hamburg, Germany.
