This study presents a Bridge Weigh-in-Motion(BWIM)method that integrates with geomagnetic positioning and strain area lateral transfer matching,which makes the vehicle signal extraction and lateral positioning easier for a multi-lane bridge under large traffic flow.This method utilizes the geomagnetic matrix vector characteristics to build up a 1-D convolution neural network(1D-CNN)to identify vehicle amount and lateral spacing between vehicles when multiple vehicles passing the bridge,subsequently,the load identification calculation method based on strain area lateral transfer matching is used to identify vehicle weights,together with the measured strain response and vehicle payload.A multi-lane hollow core beam bridge in Wuhan was used as a case.Vehicle lateral position positioning tests were performed.Calibration tests were conducted to obtain the actual influence surfaces of the bridge by the dynamic strain response calculation.Tests were carried out in two scenarios of one vehicle passing through the bridge and multiple vehicles passing through the bridge;Based on the cross-correlation algorithm,the vehicle velocity is inverted,furthermore,the vehicle weight is identified,and the identification accuracy is analyzed.It is demonstrated that the presented method can accurately identify the vehicle amount and lateral positions,with a lateral position identification accuracy maintained at about 96％,vehicle velocity identification error below 5％.In the scenario of single vehicle passing,the vehicle weight identification error is below 2％,while below 5％in the multiple vehicles passing scenario.The method presented can remarkably reduce the calculation complexities and disruption of dynamic response,with ensured the vehicle weight identification accuracy,which is also applicable in the scenario of multiple vehicles passing.
关键词
桥梁工程/车辆荷载识别/地磁定位方法/BWIM系统/多车道/车辆重量/横向位置/现场试验
Key words
bridge engineering/vehicle load identification/geomagnetic positioning method/BWIM system/multi-lane/vehicle weight/lateral position/field test
维普
万方数据