基于多视角信息融合的公路桥梁车辆荷载时空位置识别  

Identification of the spatiotemporal location of vehicle loads on highway bridges based on multi-view information fusion

作　　者：邓露[1,2] 邓佳宇 王维[1] 何维[1] 张龙威 Deng Lu;Deng Jiayu;Wang Wei;He Wei;Zhang Longwei(College of Civil Engineering,Hunan University,Changsha 410082,China;Hunan Provincial Key Laboratory for Damage Diagnosis of Engineering Structures,Hunan University,Changsha 410082,China;College of Civil Engineering,Hunan University of Science and Technology,Xiangtan 411201,China)

机构地区：[1]湖南大学土木工程学院,长沙410082 [2]湖南大学工程结构损伤诊断湖南省重点实验室,长沙410082 [3]湖南科技大学土木工程学院,湘潭411201

出　　处：《Journal of Southeast University(English Edition)》2024年第1期1-12,共12页东南大学学报（英文版）

基　　金：The National Natural Science Foundation of China Youth Program(No.52108139);Hunan Provincial Natural Science Foundation Youth Program(No.2023JJ40290).

摘　　要：针对现有方法在复杂交通场景下难以准确获取桥上车辆荷载时空分布的问题,提出了一种基于多视角信息融合的车辆荷载时空位置识别方法.首先,开发了vadYOLO-StrongSORT模型,可在单视角下同时检测和跟踪车辆;然后,在图像标定和跨视角车辆匹配基础上,采用自适应加权最小二乘法进行多视角信息融合以修正车辆轨迹;最后,结合车辆轨迹和车轴配置,重构车轴荷载的时空位置分布.通过模型试验评估了所提方法在典型交通场景下的性能.结果表明:相较于基于单视角的车辆位置识别方法,多视角信息融合方法在跟踪稳定性、定位精度和抗遮挡性能上有显著提升;变道场景下,所提方法的最高平均定位误差低于2.0 cm,明显优于单视角方法的17.0 cm;多车遮挡场景下,所提方法的车辆捕获率可达100%,而单视角方法最高仅为72.5%;同时,与其他检测跟踪模型相比,vadYOLO-StrongSORT在试验中取得了最高的识别精度.To solve the problem that existing methods have difficulty in accurately obtaining the spatiotemporal distribution of vehicle loads on bridges in complicated traffic scenes,a spatiotemporal location identification method for vehicle loads based on multi-view information fusion is proposed.First,the vadYOLO-StrongSORT model is developed to detect and track vehicles simultaneously in a single view.Furthermore,based on image calibration and cross-view vehicle matching,an adaptive weighted least squares method is used for multi-view information fusion to correct the vehicle trajectory.Finally,the spatiotemporal distribution of axle loads is reconstructed by combining vehicle trajectories with axle configurations.The performance of the proposed method under typical traffic conditions is evaluated using model tests.The results show that the multi-view information fusion method significantly improves tracking stability,localization accuracy,and anti-occlusion performance compared with the single view-based vehicle location identification method.In the lane-changing scenes,the highest average localization error of the proposed method is less than 2.0 cm,which is significantly better than the 17.0 cm of the single-view method.In multivehicle occlusion scenes,the proposed method achieves a vehicle capture rate of up to 100%,compared with a maximum of only 72.5%for the single-view method.Meanwhile,vadYOLO-StrongSORT achieves the highest identification accuracy in the experiment compared with other detection and tracking models.

关 键 词：桥梁工程 车辆荷载 时空位置 多视角信息融合 车轴识别 桥梁动态称重 桥梁健康监测 

分 类 号：U441.3[建筑科学—桥梁与隧道工程]