Identification of Frequencies and Track Irregularities of Railway Bridges Using Vehicle Responses: A Recursive Bayesian Kalman Filter Algorithm

On-board monitoring of track irregularities and bridge dynamic characteristics based on vehicle vibration responses provides basic data for the condition assessment of high speed railway bridges. However, the identification process inevitably introduces estimation uncertainty because of measurement noise and system parameter uncertainty. Here, in a probability framework, we propose a recursive Bayesian Kalman filtering (RBKF) algorithm for quantifying the identification uncertainty of the track irregularities and bridge natural frequencies. A nonlinear state-space model with measurement noise and process noise was first established for vehicle-bridge (VB) systems. Then the RBKF algorithm was formulated using a nonlinear state-space model, and the identification uncertainty was quantified in terms of estimation variances. A numerical study of two high speed railway bridges validated the RBKF algorithm. This study may help develop new approaches for on-board monitoring and condition assessment of high speed railway bridges.