A GENERAL FRAMEWORK FOR DAMAGE DETECTION WITH STATISTICAL DISTANCE MEASURES: APPLICATION TO CIVIL ENGINEERING STRUCTURES

Detecting damage in structural systems is often achieved by a statistical comparison of damage-sensitive characteristics of a structure evaluated on baseline data, against the corresponding characteristics obtained using data collected from a potentially defective structure. While several vibration-based methods have been proposed and successfully applied to detect damage in both mechanical and civil structures over the past years, the general framework describing their common properties and unifying the statistical decision about damage has mainly been elaborated in the control community. In this paper, we revise this framework in the context of detecting damage in structural systems. The statistical properties of three commonly used damage detection methods are recalled, and it is shown that their evaluation for damage boils down to a simple statistical distance. The framework is adopted to a commercial structural health monitoring software suite and its practical merit is illustrated on damage detection of two full-scale highway bridges.