A Hybrid Model-Free Data-Interpretation Approach for Damage Detection during Continuous Civil Infrastructure Monitoring

Abstract

A hard challenge associated with infrastructure monitoring is to extract useful information from large amounts of measurement data in order to detect changes in structures. This paper presents a hybrid model-free approach that combines two model-free methods - Moving Principal Component Analysis (MPCA) and Robust Regression Analysis (RRA) - to detect damage during continuous monitoring of structures. While a merit of MPCA is the ability to detect small amount of damage, an advantage of RRA is fast damage detection. The objective of this paper is to exploit these two complementary advantages through an appropriate combination. The applicability of this hybrid approach is studied on a railway truss bridge in Zangenberg (Germany). Its performance is compared with that of individual methods in terms of damage detectability and time to detection. Results show that the hybrid approach has higher damage detectability and identifies damage faster than individual applications of MPCA and RRA.