Structural Damage Warning of a Long-Span Cable-Stayed Bridge Using Novelty Detection Technique Based on Wavelet Packet Analysis

Abstract

This paper describes the development of a multi-stage scheme for structural damage warning of the Runyang Cable-stayed Bridge using the measured dynamic responses from an on-line instrumentation system. In the first stage, the wavelet packet energy spectrum (WPES) is extracted using wavelet packet analysis from the measured dynamic responses caused by ambient excitations, thus providing parameters warning of occurrence of structural damage. The second stage is to formulate correlation models describing the seasonal relationship between WPES and temperature and the third stage is to classify the measured changes of WPES, whether due to environmental temperature or to structural damage, using a novelty detection technique based on statistical process control. The usefulness of the approach is examined on the Runyang Cable-stayed Bridge using 236 days of health monitoring data. The results reveal that the seasonal change of environmental temperature accounts for variation in the measured WPES, which averaged a 200% variance, and that the approach is able to eliminate the temperature effects and provides a good capability for detecting the damage-induced 10% variance.