A0813
Title: Functional quantile analysis for sensor outputs in structural health monitoring
Authors: Frederike Vogel - Helmut-Schmidt-University, Hamburg (Germany) [presenting]
Abstract: Structural health monitoring is a pivotal discipline in determining the condition of a given structure, e.g., a bridge, by gathering and assessing data from sensory systems attached to it. These sensor data can be interpreted as functional. As structural damage can impact the structure's service life, detecting potential damage as quickly as possible is important. A comprehensive analysis of all signals' distributions is essential to achieve this. However, conventional monitoring concepts based on, for instance, functional principal component analysis (FPCA) fall short in accounting for skewness or shifting effects as they merely represent curves as deviations from the mean. In this innovative approach, FPCA is expanded by incorporating a quantile perspective, thereby considering scores at various quantile levels as vital monitoring metrics. Furthermore, the model takes into account confounding effects, specifically the temperature. The method is validated through simulation studies and real-data scenarios.
