Dynamic Monitoring Method of Large Bridge Vehicle Loads based on Multi-target Tracking
Dynamic distribution of vehicle loads on bridges is important for monitoring and analysis of bridge health status.Traditional vehicle load monitoring methods are difficult to be applied to bridges with large spans and complex structures,as well as incapable to monitor in real time.It is therefore proposed in this paper,based on the use of weight in motion system,a machine vision-based dynamic monitoring method for vehicle loads,which can be applied to large span bridges,to address the limitations of traditional vehicle load monitoring methods.First,a weight in motion system is arranged at the bridge to collect weight data from passing vehicles,while a computer vision system is used to synchronize the images of the weighed vehicles and assign IDs.In order to calculate the position of the vehicle on the bridge during the vehicle's travel along the bridge,the image perspective distortion correction is first carried out to establish the mapping relationship between the image coordinates and the real bridge coordinates,and then the SSD algorithm is used to identify the vehicle within the field of view of the camera along the bridge through the vision detection system arranged on the bridge to save the position information of the vehicle on the bridge.Then the ID correlation of vehicle targets at adjacent moments is realized according to multi-target tracking algorithm.Iteration is repeated,and the moving track of the vehicle in the whole bridge can be obtained,and its weight information can be traced according to the vehicle ID,so as to achieve the purpose of real-time monitoring of the bridge vehicle load.Experiment validation is conducted at Wuhan Yingwuzhou Yangtze River Bridge in order to verify the reliability of the proposed method.The results show that the accuracy of vehicle identification under different weather is 96.3％and the ID variation rate of multi-target tracking results is 3.4％.The experimental results show that the monitoring effect is stable and reliable,and can be satisfied with the requirements of monitoring and recording the dynamic distribution of vehicle loads on the whole bridge surface of large span bridges.
bridge monitoringidentification of vehicle loadcomputer visiondeep learningmulti-target tracking
NETL
NSTL
万方数据
