1.School of Transportation Management,People's Public Security University of China,Beijing 100038,China
2.School of Transportation,Jilin University,Changchun 130022,China
Show less
文章历史+
Received
Published
2023-11-13
2025-08-01
Issue Date
2025-10-30
PDF (4166K)
摘要
卷积神经网络(CNN)是目前交通状态估计的深度学习算法中提取交通特征的关键模块,其对稀疏数据和多模式交通状态的计算不稳定,制约深度学习的状态估计精度。为进一步提升CNN的交通特征分析精度,本文建立了一种编码-解码的交通自适应卷积网络。首先,本文所提网络构建一种交通特征编码CNN,采用下采样操作聚合邻域交通信息,以从稀疏数据中提取有效的交通特征;其次,构建交通状态自适应重构CNN,利用提取的特征准确重构不同模式的交通状态。为表征多样化交通状态时空结构,该CNN引入先验知识引导卷积核形变。最后,实验采用长春市出租车GPS数据对算法在稀疏数据和不同情景下的交通状态估计性能进行验证。实验结果表明,与LSTM、GAN等先进算法相比,本文提出的改进的CNN算法估计精度提高了6.05 km/h RMSE,同时在不同交通情景下的估计精度仅差3.34% RMSE,能为深度学习估计交通状态提供有力的交通特征分析支撑。
Abstract
Convolutional neural networks (CNN) are the key modules for extracting traffic features in deep learning algorithms for traffic state estimation. However, their computational instability on sparse data and multi-modal traffic states limits the accuracy of deep learning for state estimation. To improve the accuracy of CNN-based traffic feature analysis, this paper proposed an encode-decode traffic adaptive convolutional network. Firstly, the proposed network constructed a traffic feature encoding CNN, which uses down-sampling operations to aggregate neighboring traffic information and extracts effective traffic features from sparse data. Secondly, a traffic state adaptive reconstruction CNN is constructed to utilize the extracted features to accurately reconstruct different patterns of traffic states. To represent diverse spatio-temporal structure of traffic states, this CNN introduced prior knowledge to guide the deformation of convolutional kernels. Finally, the performance of algorithm in traffic state estimation under sparse data and different scenarios was validated using taxi GPS data in Changchun City. Experimental results show that compared with advanced algorithms such as LSTM and GAN, the improved CNN alogorithm in this paper improves the estimation accuracy by 6.05 km/h RMSE. Besides, the proposed algorithm has an accuracy difference of only 3.34% RMSE in different traffic scenarios, can provide robust traffic festure analysis support for deep learning-based traffic state estimation.
考虑道路上有一组在驶浮动车对交通状态进行采样,采样数据最终记为二维交通状态时空观测矩阵 S,浮动车在时间间隔内经过道路位置,其采样的交通信息将作为观测矩阵 S 相应时空位置的交通状态观测值。为标记交通状态时空观测矩阵 S 中缺乏采样信息的位置,本文定义掩膜矩阵 M,当未有浮动车采样的数据时,;否则,。为表征矩阵中未观测交通状态的不确定性,本文还在输入数据中引入了随机变量。
本文选取衡量道路状态估计精度的指标包括平均绝对误差(Mean absolute error,MAE)、均方根误差(Root mean square error,RMSE)、标准化平均误差(Normalized mean square error,NMSE),这3个误差指标越小,说明估计精度越高。RMSE、MAE及NMSE的计算公式分别为:
YuJ J Q, GuJ. Real-time traffic speed estimation with graph convolutional generative autoencoder[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 20(10): 3940-3951.
[2]
ZhangL, ChenT, YuB, et al. Suburban demand responsive transit service with rental vehicles[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 22(4): 2391-2403.
[3]
LiX, WangT, XuW, et al. A novel model for designing a demand-responsive connector (DRC) transit system with consideration of users' preferred time windows[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 22(4): 2442-2451.
[4]
田婧. 知识感知下的道路交通状态重构方法研究[D]. 长春: 吉林大学交通学院, 2023.
[5]
TianJing. Research on road traffic state reconstruction method in perspective of knowledge-aware[D]. Changchun: School of Transportation, Jilin University, 2023.
[6]
AhmedM S, CookA R. Analysis of freeway traffic time-series data by using box-jenkins techniques[J]. Transportation Research Record, 1979, 722: 1-9.
[7]
WilliamsB M, HoelL A. Modeling and forecasting vehicular traffic flow as a seasonal ARIMA process: Theoretical basis and empirical results[J]. Journal of Transportation Engineering, 2003, 129(6): 664-672.
ChenXi-qun, CaoZhen, MoDong. Analyzing error bounds of highway traffic state estimation via Kalman filter fusion[J]. Journal of Transportation Systems Engineering and Information Technology, 2022, 22(4): 72-78.
[10]
TangJ, ZhangG, WangY, et al. A hybrid approach to integrate fuzzy C-means based imputation method with genetic algorithm for missing traffic volume data estimation[J]. Transportation Research Part C: Emerging Technologies, 2015, 51(4): 29-40.
ChenJia-liang, HuZhao-zheng, LiFei. An estimation method of traffic flow state based on matching of temporal-spatial feature sequences[J]. Journal of Transport Information and Safety, 2021, 39(3): 68-76.
[13]
XiaoJ, WeiC, LiuY. Speed estimation of traffic flow using multiple kernel support vector regression[J]. Physica A: Statistical Mechanics and its Applications, 2018, 509(11): 989-997.
[14]
ChenX, HeZ, SunL. A bayesian tensor decomposition approach for spatiotemporal traffic data imputation[J]. Transportation Research Part C: Emerging Technologies, 2019, 98: 73-84.
[15]
TianY, ZhangK, LiJ, et al. LSTM-based traffic flow prediction with missing data[J]. Neurocomputing (Amsterdam), 2018, 318: 297-305.
[16]
ZhangS, ZhouL, ChenX, et al. Network‐wide traffic speed forecasting: 3D convolutional neural network with ensemble empirical mode decomposition[J]. Computer-Aided Civil and Infrastructure Engineering, 2020, 35(10): 1132-1147.
[17]
RempeF, FraneckP, BogenbergerK. On the estimation of traffic speeds with deep convolutional neural networks given probe data[J]. Transportation Research Part C: Emerging Technologies, 2022, 134(1): 103448.
[18]
OuafaB, BilalT T, HwasooY, et al. Traffic data imputation using deep convolutional neural networks[J]. IEEE Access, 2020, 8: 104740-104752.
[19]
DaiJ, QiH, XiongY, et al. Deformable convolutional networks[C]∥Proceeding of the IEEE International Conference on Computer Vision. Los Alamitos: IEEE Computer Society, 2017: 764-773.
[20]
ZhuX, HuH, LinS, et al. Deformable convnets V2: More deformable, better results[C]∥Proceeding of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Los Alamitos: IEEE Computer Society, 2020: 9300-9308.
[21]
DoerschC. Tutorial on variational autoencoders[J]. Arxiv Preprint, 2016, 6: 160605908.
[22]
DilipD M, FrerisN M, JabariS E. Sparse estimation of travel time distributions using Gamma kernels[C]∥Transportation Research Board 96th Annual Meeting. Washington: Transportation Research Board, 2017: 01628670.
[23]
JabariS E G, DilipD M, LinD, et al. Learning traffic flow dynamics using random fields[J]. IEEE Access, 2019, 7: 130566-130577.
[24]
TreiberM, KestingA, WilsonR E. Reconstructing the traffic state by fusion of heterogeneous data[J]. Computer-Aided Civil and Infrastructure Engineering, 2011, 26(6): 408-419.
[25]
SchererD, MüllerA, BehnkeS. Evaluation of pooling operations in convolutional architectures for object recognition[C]∥Lecture Notes in Computer Science. Berlin: Springer, 2010: 92-101.
[26]
TanC P, YaoJ R, TangK S, et al. Cycle-based queue length estimation for signalized intersections using sparse vehicle trajectory data[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 22(1): 91-106.
[27]
MasciJ, MeierU, CiresanD, et al. Stacked convolutional auto-encoders for hierarchical feature extraction[C]∥Proceeding of the International Conference on Artificial Neural Networks. Berlin: Springer, 2011: 52-59.
[28]
ChenX, YangJ, SunL. A nonconvex low-rank tensor completion model for spatiotemporal traffic data imputation[J]. Transportation Research Part C: Emerging Technologies, 2020, 117(8): 102673.
[29]
CuiZ, KeR, PuZ, et al. Stacked bidirectional and unidirectional LSTM recurrent neural network for forecasting network-wide traffic state with missing values[J]. Transportation Research Part C: Emerging Technologies, 2020, 118(9): 102674.
[30]
ChenY, LvY, WangF Y. Traffic flow imputation using parallel data and generative adversarial networks[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, 21(4): 1624-1630.
[31]
ZhangK, HeZ, ZhengL, et al. A generative adversarial network for travel times imputation using trajectory data[J]. Computer-Aided Civil and Infrastructure Engineering, 2021, 36(2): 197-212.