Long-term bridge health monitoring focusing on the Mahalanobis Distance of modal parameters

Abstract

Maintaining civil infrastructure, including bridges, has been a keen technical issue in developed countries and will surely be one in developing countries in the near future. An effective maintenance strategy strongly depends on timely decisions on the health condition of the structure. Bridge health monitoring (BHM) using vibration data is widely recognized to be one of the effective technologies that aid decision-making on bridge maintenance. This research focuses on long-term BHM expecting that changes in physical properties of the bridge subject to aged-deterioration progress slowly. In the practical application of the long-term BHM, one of the difficulties is that the observed vibration data includes environmental influences such as temperature change. In order to achieve high accuracy in evaluating modal parameters of the bridge, other influencing variables have to be taken into consideration. In this study, temperature is considered as the main environmental factor by means of a regression analysis. The Mahalanobis distance (MD), a multivariate statistical distance, is adopted to emphasize potential changes in the identified modal parameters. The validity of the proposed approach is investigated utilizing vibration data measured at a real bridge in service.