AbstractStructural health monitoring of bridges based on train-induced vibrations has attracted extensive attention and takes advantage of the high-magnitude and regularly spaced features of trainloads. However, the monitoring process is restricted because the relationship between bridge characteristics and the dynamic features of train-induced vibration responses is not well represented. In this study, train-induced acceleration characteristics of a bridge are investigated, and the relationship between the train speed and spectral amplitude of the bridge acceleration at a frequency equal to the ratio of the speed to carriage length is revealed. Based on the phenomenon that the speed–amplitude correlation trend is consistent with the structural frequency response function (FRF), a dynamic index called the equivalent frequency response function (EFRF) is proposed. The evolution rules of the EFRF curve with varying natural frequencies and train loads are distinguished, and the structural performance variation can be determined according to the relationship between the structural stiffness, natural frequencies, and EFRF. Because the speed–amplitude points are scattered due to the train configuration parameters and measurement noise, the EFRF curve should be fitted first. In addition, the normal fluctuation in the speed–amplitude points is limited using the local Shewhart control chart, considering the heteroscedasticity of the speed–amplitude points. Then, speed–amplitude points that exceed the control limits are used to identify abnormal train-induced vibrations. Finally, the monitoring data for a railway bridge with multiple lanes are considered to verify the proposed method. The results show that the abnormal train-induced vibration warning based on speed–amplitude points have a similar performance to that based on the speed–acceleration correlation, but the causes of the abnormal vibration can be identified explicitly only with the speed–amplitude correlation.

Source link

Leave a Reply

Your email address will not be published.