Eliminating the Bridge Modal Variability Induced by Thermal Effects Using Localized Modeling Method

Abstract

Eliminating the bridge modal variability induced by temperature interference is crucial in vibration-based damage detection. In this paper, a novel localized thermal–frequency correlation model is proposed to remove the masking effect. Motivated by the modal frequencies may be generated and affected by different temperature mechanisms and nonlinear dependence exists between them, a Gaussian mixture model was introduced to cluster the temperature and frequency data into different subsets. Then, a new representative temperature of bridge temperature field was deduced through partial least squares, which is sensitive to modal frequency. After that, the more accurate baseline correlation models were established. The performance alarming was implemented combining with kernel density estimate, which deals with the modeling error nonnormal distribution, and X-bar control chart. Finally, an engineering application to a cable-stayed bridge was carried out, which demonstrates the validity of eliminating the temperature-induced change in modal frequency based on long-term monitoring data.