基于动态生成对抗网络的路网缺失交通数据修复

doi:10.16088/j.issn.1001-6600.2023051402

摘要/Abstract

摘要： 针对智能交通系统数据采集过程中发生的数据缺失问题,本文提出一种基于动态生成对抗网络(dynamic generative adversarial network, D-GAN)的路网交通数据修复方法。该方法首先依据交通数据的时空特性与设定的缺失类型和缺失比例来构造各种缺失交通数据矩阵,然后基于博弈思想迭代训练由2个全连接神经网络构成的生成对抗网络。引入一种新颖的动态自适应机制,研究能在模型计算过程中自动识别生成器与判别器的最佳迭代次数,最终生成完整的交通数据矩阵并修复缺失值。采用加州PeMS和广州交通速度数据集来完成D-GAN模型的构建,并使用多种评价指标评估D-GAN的修复性能。实验结果表明:相对于非随机缺失模式,D-GAN对随机缺失模式的修复精度更高;随着缺失率增加,D-GAN的修复精度加速下降。但在各种缺失条件下,D-GAN模型的修复性能要优于现有模型(例如BGCP、prophet-RF和GAIN)。

关键词: 智能交通系统, 交通数据修复, 生成对抗网络, 博弈思想, 动态自适应机制

Abstract: For the issue of missing data occurring in the data collection process of intelligent transportation systems, a road network traffic data recovery model is proposed in this paper based on the dynamic generative adversarial network. Firstly, the approach in this paper are constructed various missing traffic data matrices by considering the spatial-temporal properties of traffic data and the set missing patterns and missing rates. Then iteratively trains a GAN was composed of two fully connected neural networks based on a game idea. By introducing a novel dynamic adaptive mechanism, this study can automatically identify the optimal number of iterations of the generator and discriminator during the model computation, and finally generates the complete traffic data matrix and repair the missing values. California PeMS and Guangzhou traffic speed datasets are used to complete the D-GAN model construction, and multiple evaluation metrics are employed to assess the repair performance of D-GAN. Experimental results show that the repair accuracy of D-GAN is higher for random missing patterns compared with non-random missing patterns; and the repair accuracy of D-GAN degrades accelerated with increasing missing rates. However, the repair performance of D-GAN outperforms the baseline models (e.g., BGCP, prophet-RF, and GAIN) under various missing conditions.

Key words: intelligent transportation systems, traffic data recovery, generative adversarial network, game idea, dynamic adaptive mechanism

中图分类号: U495

许伦辉, 李金龙, 李若南, 陈俊宇. 基于动态生成对抗网络的路网缺失交通数据修复[J]. 广西师范大学学报（自然科学版）, 2024, 42(2): 30-40.

XU Lunhui, LI Jinlong, LI Ruonan, CHEN Junyu. Missing Traffic Data Recovery for Road Network Based on Dynamic Generative Adversarial Network[J]. Journal of Guangxi Normal University(Natural Science Edition), 2024, 42(2): 30-40.

参考文献

[1] LI H P, LI M, LIN X, et al. A spatiotemporal approach for traffic data imputation with complicated missing patterns[J]. Transportation Research Part C: Emerging Technologies, 2020, 119: 102730. DOI:10.1016/j.trc.2020.102730.
[2] LI J L, WU P, GUO H C, et al. Multivariate transfer passenger flow forecasting with data imputation by joint deep learning and matrix factorization[J]. Applied Sciences, 2023, 13(9): 5625. DOI:10.3390/app13095625.
[3] LI Y B, LI Z H, LI L. Missing traffic data: comparison of imputation methods[J]. IET Intelligent Transport Systems, 2014, 8(1): 51-57. DOI:10.1049/iet-its.2013.0052.
[4] LI J L, WU P, LI R N, et al. ST-CRMF: compensated residual matrix factorization with spatial-temporal regularization for graph-based time series forecasting[J]. Sensors, 2022, 22(15): 5877. DOI:10.3390/s22155877.
[5] MA X L, LUAN S, DING C, et al. Spatial interpolation of missing annual average daily traffic data using copula-based model[J]. IEEE Intelligent Transportation Systems Magazine, 2019, 11(3): 158-170. DOI:10.1109/MITS.2019.2919504.
[6] HAN T Y, WADA K, OGUCHI T. Large-scale traffic data imputation using matrix completion on graphs[C]// 2019 IEEE Intelligent Transportation Systems Conference (ITSC). Piscataway, NJ: IEEE, 2019: 2252-2258. DOI:10.1109/ITSC.2019.8917365.
[7] LI J L, XU L H, LI R N, et al. Deep spatial-temporal bi-directional residual optimisation based on tensor decomposition for traffic data imputation on urban road network[J]. Applied Intelligence, 2022, 52(10): 11363-11381. DOI:10.1007/s10489-021-03060-4.
[8] REN P, CHEN X Y, SUN L J, et al. Incremental Bayesian matrix/tensor learning for structural monitoring data imputation and response forecasting[J]. Mechanical Systems and Signal Processing, 2021, 158: 107734. DOI:10.1016/j.ymssp.2021.107734.
[9] 陆文琦,周天,谷远利,等.基于张量分解理论的车道级交通流数据修复算法[J].吉林大学学报(工学版),2021,51(5): 1708-1715.DOI:10.13229/j.cnki.jdxbgxb20200535.
[10] 武江南,张红梅,赵永梅,等.基于张量奇异值理论的交通数据重构方法[J].计算机应用研究,2022,39(5): 1449-1453,1459.DOI:10.19734/j.issn.1001-3695.2021.10.0429.
[11] CHEN X Y, LEI M Y, SAUNIER N, et al. Low-rank autoregressive tensor completion for spatiotemporal traffic data imputation[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(8): 12301-12310. DOI:10.1109/TITS.2021.3113608.
[12] RAN B, TAN H C, WU Y K, et al. Tensor based missing traffic data completion with spatial-temporal correlation[J].Physica A: Statistical Mechanics and Its Applications, 2016, 446: 54-63. DOI:10.1016/j.physa.2015.09.105.
[13] ROLL J. Daily traffic count imputation for bicycle and pedestrian traffic: comparing existing methods with machine learning approaches[J]. Transportation Research Record, 2021, 2675(11): 1428-1440. DOI:10.1177/03611981211027161.
[14] ZHANG Y, LIU Y C. Data imputation using least squares support vector machines in urban arterial streets[J]. IEEE Signal Processing Letters, 2009, 16(5): 414-417. DOI:10.1109/LSP.2009.2016451.
[15] 裴莉莉,孙朝云,韩雨希,等.基于SSC与XGBoost的高速公路异常收费数据修复算法[J].吉林大学学报(工学版),2022,52(10): 2325-2332.DOI:10.13229/j.cnki.jdxbgxb20210231.
[16] TANG J J, CHEN X Q, HU Z, et al. Traffic flow prediction based on combination of support vector machine and data denoising schemes[J]. Physica A: Statistical Mechanics and Its Applications, 2019, 534: 120642. DOI:10.1016/j.physa.2019.03.007.
[17] 李振亮,李波.基于矩阵分解的卷积神经网络改进方法[J].计算机应用,2023,43(3): 685-691.DOI:10.11772/j.issn.1001-9081.2022010032.
[18] YANG J M, PENG Z R, LIN L. Real-time spatiotemporal prediction and imputation of traffic status based on LSTM and Graph Laplacian regularized matrix factorization[J]. Transportation Research Part C: Emerging Technologies, 2021, 129: 103228. DOI:10.1016/j.trc.2021.103228.
[19] 钱斌,郑楷洪,陈子鹏,等.基于残差连接长短期记忆网络的时间序列修复模型[J].计算机应用,2021,41(1): 243-248.DOI:10.11772/j.issn.1001-9081.2020060928.
[20] ZHANG Z C, LIN X, LI M, et al. A customized deep learning approach to integrate network-scale online traffic data imputation and prediction[J]. Transportation Research Part C: Emerging Technologies, 2021, 132: 103372. DOI:10.1016/j.trc.2021.103372.
[21] GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial networks[J]. Communications of the ACM, 2020, 63(11): 139-144. DOI:10.1145/3422622.
[22] 徐东伟,彭航,商学天,等.基于图自编码-生成对抗网络的路网数据修复[J].交通运输系统工程与信息,2021,21(6): 33-41.DOI:10.16097/j.cnki.1009-6744.2021.06.005.
[23] ZHANG S C. Nearest neighbor selection for iterativelykNN imputation[J]. Journal of Systems and Software, 2012, 85(11): 2541-2552. DOI:10.1016/j.jss.2012.05.073.
[24] CHAUDHRY A, LI W, BASRI A, et al. A method for improving imputation and prediction accuracy of highly seasonal univariate data with large periods of missingness[J]. Wireless Communications and Mobile Computing, 2019,2019: 4039758. DOI:10.1155/2019/4039758.
[25] CHEN X Y, HE Z C, SUN L J. A Bayesian tensor decomposition approach for spatiotemporal traffic data imputation[J]. Transportation Research Part C: Emerging Technologies, 2019, 98: 73-84. DOI:10.1016/j.trc.2018.11.003.
[26] YOON J, JORDON J, SCHAAR M. GAIN: missing data imputation using generative adversarial nets[J]. Proceedings of Machine Learning Research, 2018, 80: 5689-5698.
[27] LI J L, SUN L J, LI R N, et al. Application ofsiSVR-Vis/NIR to the nondestructive determination of acid detergent fiber content in corn straw[J]. Optik, 2020, 202: 163717. DOI:10.1016/j.ijleo.2019.163717.
[28] LI R N, QIN Y, WANG J B, et al. AMGB: trajectory prediction using attention-based mechanism GCN-BiLSTM in IOV[J]. Pattern Recognition Letters, 2023, 169: 17-27. DOI:10.1016/j.patrec.2023.03.006.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed