Health condition monitoring of civil structures using time varying autoregressive models

Abstract

In recent years, there have been an increasing interest in long-term monitoring of civil structures, as the research community has been alarmed by some tragic events and collapses of bridges and buildings that pointed out the vulnerability of some existing structures and the uncertainties in their analysis for monitoring and maintenance purposes. SHM is the measurement of the operating and loading environment; as well as the critical responses of a structure to track and evaluate the symptoms of incidents, anomalies, damage and/or deterioration which may affect operation, serviceability, safety and reliability. Although many damage detection techniques were applied to scaled models or specimen tests in controlled laboratory environments, the performance of these techniques in real operational environments is still questionable and needs to be validated. Often damage sensitive features employed in these damage detection techniques are also sensitive to changes of environmental and operation conditions of the structure. The objective of this study is to propose a new Time Varying Autoregressive (TVAR) modeling technique for SHM of large-scale structures like bridges and buildings. TVAR model, a method by virtue of its nature is applicable for modeling data whose spectral content varies with time. The research is conducted to critically understand the effective performance of the structures under various loads and health conditions, and detect their operational anomalies using the proposed data-driven technique. In this research, an attempt is made to alleviate the use of system identification method where TVAR modeling is conducted directly on the data. The proposed method does not depend on the complicated algorithms and free of any other user-defined parameters. In pursuance of applying the proposed data-driven technique, the data collected on site are essentially paramount. Data inherently used are mainly obtained from experiments, as well as the data acquired from the Harbin Institute of Technology in fulfillment of a full-scale validation. The proposed TVAR technique detects not only the occurrence of structural damage, but also the location of damage. Whereas the TVAR developed captures the changes in the time domain, for comparison, Stochastic Subspace System Identification (SSI) method is applied to the experimental data. The method is used because it is an important tool that captures the frequency changes, as the SSI tracks the changes in the frequency domain. Using both experimental and full-scale studies, it is shown that the proposed TVAR technique and the comparable SSI method applied, can therefore be considered as a useful tool for SHM.