融合多种算法的车辆轨迹重构框架研究  

Research on Vehicle Trajectory Reconstruction Framework Integrating Multiple Algorithms

作　　者：龙浩天 赵丹[1] LONG Haotian;ZHAO Dan(People s Public Security University of China,Beijing 100091,China)

机构地区：[1]中国人民公安大学,北京100091

出　　处：《交通工程》2025年第5期84-89,95,共7页Journal of Transportation Engineering

基　　金：“自主式道路交通系统安全保障技术”(2023YFB4302703)。

摘　　要：为提高轨迹数据的准确性,提出一种融合多种算法的车辆轨迹三步重构框架:首先,利用小波变换识别并修正轨迹数据中的异常值;其次,应用基于高斯核的局部加权线性回归对数据进行插值处理;最后,通过卡尔曼滤波对轨迹数据进行平滑处理。通过对NGSIM车辆轨迹数据库中的I-80数据集进行实验验证,结果表明该重构框架能降低轨迹数据中的异常值比例、平滑效果显著,减少加加速度(Jerk值)的极端波动,提高数据的内部和物理一致性。本研究提出的三步车辆轨迹重构框架在处理轨迹数据中的异常值和测量误差方面表现出色,为交通工程领域提供了一种数据处理方法。To enhance the accuracy of trajectory data,this study proposes a three-step vehicle trajectory reconstruction framework that integrates multiple algorithms.First,wavelet transform is employed to identify and correct outliers in the trajectory data.Second,local weighted linear regression based on a Gaussian kernel is applied for data interpolation.Finally,Kalman Filtering is utilized to smooth the trajectory data.The effectiveness of the proposed framework was validated using the I-80 dataset from the NGSIM vehicle trajectory database.The results demonstrate that the reconstruction framework significantly reduces the proportion of outliers in the trajectory data,achieves notable smoothing effects,and effectively mitigates extreme fluctuations in jerk values(i.e.,the rate of change of acceleration),thereby improving the internal and physical consistency of the data.The three-step vehicle trajectory reconstruction framework proposed in this study performs exceptionally well in handling outliers and measurement errors in trajectory data,providing an effective data processing method for the field of traffic engineering.

关 键 词：轨迹重构 小波变换 局部加权回归 卡尔曼滤波 

分 类 号：F321[经济管理—产业经济]