Probabilistic Structural Health Monitoring Method Applied to the Bridge Health Monitoring Benchmark Problem

Abstract

A probabilistic structural health monitoring (SHM) method was identified to detect global damage in a highway bridge while accounting for the variability associated in calculating the desired damage measure. The proposed vibration-based monitoring method is applied to the bridge health monitoring benchmark problem. The benchmark structure is a scaled laboratory model of a typical highway bridge. The associated finite element model of this structure is excited with a simulated traffic excitation. The proposed SHM method is intended to meet the practical requirements of using a small number of sensors and detection of damage in a global sense. As such, accelerations are measured at a single location on the bridge. The bridge natural frequencies for a healthy state and a variety of damaged states are considered to help quantify and detect the potential damage. The natural frequencies are determined from the peaks in the power spectral density functions of the measured acceleration response. These natural frequencies are observed to be random variables with a normal distribution. The probability that the current natural frequency is greater than the previously measured healthy natural frequency can then be calculated from the assumed probability density functions and is used to predict, in a global sense, whether damage has occurred on the bridge structure. The results of this simulation-based study were used to extend and apply this method to actual highway bridges in Connecticut.