Degradation Rate Prediction of Bridges Using Historical Vibration Data

Abstract

Abstract Prediction of the future health state of civil structures allows maintenance agencies to undertake timely repair and replacement activities. Appropriate conduct of maintenance actions ensures accident-free operation of the transport network. Precise estimation of remaining useful life of civil structures is achievable through the availability of the degradation history, selection of appropriate degradation models, and efficient prognostic algorithms. In the reported research work, the variation in modal energy due to progressive degradation of the structure is used as a degradation quantification feature. Then, a sequential Monte Carlo–based particle filter (PF)–based scheme is reformulated to undertake bridge health prognosis of an aging in-service concrete bridge situated in a harbor area, using historical nondestructive testing data. A microelectromechanical systems–based accelerometer sensors are installed at different bridge segments to record vibration data. The historical database of the degradation feature is segmented into training and validation regions. The particle filter algorithm ultimately predicts the posterior probability density function of the degradation quantification feature for the next time instant(s). The promising prognostic results highlight the efficacy of the proposed scheme.