Description
High-speed railway bridges are subjected to heavy operational loads and significant impact forces due to the moving trains. It is important to carry out an effective structural health monitoring to ensure the safe operation of railway bridges. Traditional online bridge monitoring methods require sensor installations on the structures that can be time-consuming and expensive making it very difficult to satisfy the huge needs of monitoring all the existing railway bridges. Therefore, this study proposes to use the drive-by method for the bridge modal identification using responses of in-service high-speed trains. The measuring sensors are installed on the train that can conduct the measurement in its normal operational states. To improve the feasibility and accuracy of drive-by bridge modal identification, the contact-point (CP) responses between the wheels and the rail track are identified first from the train responses. A novel method via Bayesian expectation-maximization based augmented Kalman filter is proposed to reconstruction the CP responses and unknown states of the train simultaneously. The method is robust to the measurement noise of the train responses and the CP responses can be identified accurately. The identified CP responses successfully eliminate the dynamic components related to the train and greatly enhance the bridge dynamic information. The proposed CP response reconstruction method has great potential for drive-by bridge modal identification using the responses of train that can be further used to assess the structural condition of high-speed railway bridges.
