基于有向图卷积与门控循环单元的短时交通流预测方法

崔文岳; 谷远利; 赵胜利; 芮小平

doi:10.3963/j.jssn.1674-4861.2023.02.013

基于有向图卷积与门控循环单元的短时交通流预测方法

doi: 10.3963/j.jssn.1674-4861.2023.02.013

1.
北京交通大学综合交通运输大数据应用技术交通运输行业重点实验室北京 100044
2.
山西太行云顶文化旅游发展有限公司山西晋城 048000
3.
河海大学地球科学与工程学院南京 211100

基金项目:

国家自然科学基金项目 41771478

北京市科技计划项目 Z121100000312101

详细信息

作者简介:
崔文岳（1998—），硕士研究生. 研究方向：交通规划与管理. E-mail: cui_wen_yue@163.com

通讯作者:
芮小平（1975—），博士，教授. 研究方向：交通地理信息系统理论与应用研究等. E-mail: ruixpsz@163.com

中图分类号: U491.1+4
计量
- 文章访问数: 713
- HTML全文浏览量: 266
- PDF下载量: 33
- 被引次数: 0
出版历程
- 收稿日期: 2022-02-22
- 网络出版日期: 2023-06-19

A Method of Predicting Short-term Traffic Flows Based on a DGC-GRU Model

1.
Key Laboratory of Transport Industry of Big Data Application Technologies for Comprehensive Transport, Beijing Jiaotong University, Beijing 100044, China
2.
Shanxi Taihangyunding Cultural Tourism Development Corporation, Jincheng 048000, Shanxi, China
3.
School of Earth Sciences and Engineering, Hohai University, Nanjing 211000, China

摘要

摘要: 为了充分挖掘快速路交通流时空特性，解决当前城市快速路交通流预测存在交通流时空特性挖掘不充分等问题，进一步提高城市快速路短时交通流的预测精度与效率，研究了基于有向图卷积神经网络和门控循环单元的组合模型（directed graph convolution network-gate recurrent unit，DGC-GRU）的城市快速路短时交通流预测方法。该方法提出空间相关性矩阵并将其引入图卷积神经网络中，构建有向图卷积神经网络用于表征交通流的有向性和流动性。将交通流参数输入有向图卷积神经网络后得到有向图卷积算子，并将有向图卷积算子引入门控循环单元，通过有向图卷积神经网络捕捉交通流的空间特性，通过门控循环单元捕捉交通流的时间特性，输出快速路交通流预测结果。选取西雅图环形快速路感应器检测数据进行实例分析，对比模型预测效果。结果表明：在数据集与参数设置均相同的情况下，DGC-GRU交通流预测模型的训练收敛速度更快，且平均绝对误差（mean absolute error，MAE）和平均绝对百分比误差（mean absolute percentage error，MAPE）均优于对比模型，与传统的GRU、GCN、DGC-LSTM模型相比，DGC-GRU模型能够将MAE和MAPE指数分别降低33.01%、5.76%、1.32%和27.75%、1.15%、7.76%，表明DGC-GRU交通流预测模型能够有效挖掘城市快速路网中的交通流时空分布特征，具有良好的预测精度与效率。
- 城市快速路 /
- 短时交通流预测 /
- 深度学习 /
- 图卷积神经网络 /
- 门控循环单元
Abstract: In order to study the spatiotemporal characteristics of traffic flows on urban expressways which have not been fully explored in previous studies, a shortterm prediction method for traffic flow on urban expressways is proposed, in order to improve the prediction accuracy and efficiency based on a combined model of directed graph convolutional neural networks and gated recurrent units (DGC-GRU). The proposed method uses a spatial correlation matrix, which is also combined with a graph convolutional neural network. A directed graph convolutional neural network (DG-CNN) is developed to characterize the directionality and variability of traffic flows. Traffic flow parameters are input into the DG-CNN to obtain the directed graph convolution operator, and the directed graph convolution operator is introduced into the gated loop unit. The DG-CNN and the gated loop unit are used to capture spatial and temporal features of traffic flow, respectively, and are combined to predict traffic flow on expressways. Traffic flow data collected from a Ring Expressway of the City of Seattle is used for experiment analysis, in order to compare the performance of the proposed prediction models. Study results show that the convergence speed of the proposed DGC-GRU model is faster than other baseline models, and the mean absolute error (MAE) and mean absolute percentage error (MAPE) of the DGC-GRU model are smaller than those of the baseline models, given the same dataset and parameter settings. Compared with traditional GRU, GCN, and DGC-LSTM models, the DGC-GRU model reduces the MAE by 33.01%, 5.76%, 1.32%, and MAPE by 27.75%, 11.15%, 7.76%, respectively, which indicate that the DGC-GRU model can effectively study the spatiotemporal characteristics of traffic flows of the urban expressway network and has a better performance on prediction accuracy and efficiency than the compared models.
- urban expressway /
- short-term traffic flow forecasting /
- deep learning /
- graph convolution neural net-work /
- gate recurrent unit

HTML全文

图 1 图卷积神经网络结构图

Figure 1. Structure of graph convolutional neural networks

下载: 全尺寸图片幻灯片

图 2 LSTM与GRU结构对比图

Figure 2. Structure comparison between LSTM and GRU

下载: 全尺寸图片幻灯片

图 3 交通路网图

Figure 3. Traffic network

下载: 全尺寸图片幻灯片

图 4 DGC-GRU模型结构图

Figure 4. Structure of DGC-GRU model

下载: 全尺寸图片幻灯片

图 5 西雅图环形快速路感应器检测数据集

Figure 5. Seattle Inductive Loop Detector Dataset

下载: 全尺寸图片幻灯片

图 6 各模型的训练过程loss值迭代图

Figure 6. Iterative graph of loss value in training process of each model

下载: 全尺寸图片幻灯片

图 7 DGC-GRU交通流预测模型预测效果图

Figure 7. Prediction efficiency of DGC-GRU model

下载: 全尺寸图片幻灯片

表 1 4种预测模型的预测结果对比

Table 1. Comparison of the prediction results of the four prediction models

指标	模型
指标	GRU	DGC	DGC-LSTM	DGC-GRU
RMSE/(km/h)	16.268	8.946	8.125	7.791
MAE/(km/h)	6.229	4.428	4.229	4.173
MAPE/%	8.516	6.925	6.671	6.153

下载: 导出CSV

参考文献(23)

[1]	陈喜群, 周凌霄, 曹震. 基于图卷积网络的路网短时交通流预测研究[J]. 交通运输系统工程与信息, 2020, 20(4): 49-55. doi: 10.16097/j.cnki.1009-6744.2020.04.008 CHEN X Q, ZHOU L X, CAO Z. Shortterm network-wide traffic prediction based on graph convolutional network[J]. Journal of Transportation Systems Engineering and Information Technology, 2020, 20(4): 49-55. (in Chinese) doi: 10.16097/j.cnki.1009-6744.2020.04.008
[2]	ZHOU T, HAN G Q, XU X M, et al. A learningbased multimodel integrated framework for dynamic traffic flow forecasting[J]. Neural Processing Letters, 2019, 49(1): 407-430. doi: 10.1007/s11063-018-9804-x
[3]	ZENG B, LI C. Improved multi-variable grey forecasting model with a dynamic backgroundvalue coefficient and its application[J]. Computers & Industrial Engineering, 2018, 118(4): 278-290.
[4]	高洪波, 张登银. 基于分形与三次指数平滑的交通流量预测模型[J]. 南京邮电大学学报(自然科学版), 2018, 38(6): 63-67. https://www.cnki.com.cn/Article/CJFDTOTAL-NJYD201806014.htm GAO H B, ZHANG D Y. Traffic flow forecasting model based on fractal and cubic exponential smoothing[J]. Journal of Nanjing University of Posts and Telecommunications(Natural Science Edition), 2018, 38(6): 63-67. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-NJYD201806014.htm
[5]	ZHOU T, JIANG D Z, LIN Z Z, et al. Hybrid dual Kalman filtering model for shortterm traffic flow forecasting[J]. IET Intelligent Transport Systems, 2019, 13(6): 1023-1032. doi: 10.1049/iet-its.2018.5385
[6]	DING C, DUAN J X, ZHANG Y R, et al. Using an ARI-MA-GARCH modeling approach to improve subway shortterm ridership forecasting accounting for dynamic vola-tility[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 19(4): 1054-1064. doi: 10.1109/TITS.2017.2711046
[7]	肖宇, 赵建有, 叱干都, 等. 基于XGBoost的短时出租车速度预测模型[J]. 交通信息与安全, 2022, 40(3): 163-170. doi: 10.3963/j.jssn.1674-4861.2022.03.017 XIAO Y, ZHAO J Y, CHI G D, et al. A shortterm prediction model for taxi speed based on XGBoost[J]. Journal of Transport Information and Safety, 2022, 40(3): 163-170. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2022.03.017
[8]	CHENG S F, LU F, PENG P, et al. Short-term traffic forecast-ing: An adaptive ST-KNN model that considers spatial heterogeneity[J]. Computers, Environment and Urban Systems, 2018, 71(9): 186-198.
[9]	傅成红, 杨书敏, 张阳. 改进支持向量回归机的短时交通流预测[J]. 交通运输系统工程与信息, 2019, 19(4): 130-134, 148. https://www.cnki.com.cn/Article/CJFDTOTAL-YSXT201904019.htm FU C H, YANG S M, ZHANG Y. Promoted shortterm traffic flow prediction model based on deep learning and sup-port vector regression[J]. Journal of Transportation Systems Engineering and Information Technology, 2019, 19(4): 130-134, 148. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-YSXT201904019.htm
[10]	MA X L, YU H Y, WANG Y P, et al. Largescale transportation network congestion evolution prediction using deep learning theory[J]. Plos One, 2015, 10(3): 1-17.
[11]	YU R, LI Y G, SHAHABI C, et al. Deep learning: A generic approach for extreme condition traffic forecasting[C]. The 2017 SIAM International Conference on Data Mining, Houston, United States: Society for Industrial and Applied Mathe-matics, 2017.
[12]	FU R, ZHANG Z, LI L. Using LSTM and GRU neural net-work methods for traffic flow prediction[C]. 31st Youth Academic Annual Conference of Chinese Association of Automation(YAC), Wuhan, China: IEEE, 2016.
[13]	倪庆剑, 彭文强, 张志政, 等. 基于信息增强传输的时空图神经网络交通流预测[J]. 计算机研究与发展, 2022, 59(2): 282-293. https://www.cnki.com.cn/Article/CJFDTOTAL-JFYZ202202004.htm NI Q J, PENG W Q, ZHANG Z Z, et al. Spatial-Temporal graph neural network for traffic flow prediction based on information enhanced transmission[J]. Journal of Computer Research and Development, 2022, 59(2): 282-293. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JFYZ202202004.htm
[14]	CUI Z Y, HENRICKSON K, KE R M, et al. Traffic graph convolutional recurrent neural network: A deep learning framework for networkscale traffic learning and forecasting[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, 21(11): 4883-4894.
[15]	闫旭, 范晓亮, 郑传潘, 等. 基于图卷积神经网络的城市交通态势预测算法[J]. 浙江大学学报(工学版), 2020, 54(6): 1147-1155. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDZC202006011.htm YAN X, FAN X L, ZHENG C P, et al. Urban traffic flow prediction algorithm based on graph convolutional neural networks[J]. Journal of Zhejiang University(Engineering Sci-ence), 2020, 54(6): 1147-1155. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDZC202006011.htm
[16]	谷振宇, 陈聪, 郑家佳, 等. 基于时空图卷积循环神经网络的交通流预测[J]. 控制与决策, 2022, 37(3): 645-653. https://www.cnki.com.cn/Article/CJFDTOTAL-KZYC202203012.htm GU Z Y, CHEN C, ZHENG J J, et al. Traffic flow prediction based on STG-CRNN[J]. Control and Decision, 2022, 37(3): 645-653. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-KZYC202203012.htm
[17]	SENG D W, LU F S, LIANG Z Y, et al. Forecasting traffic flows in irregular regions with multi-graph convolutional network and gated recurrent unit[J]. Frontiers of Information Technology & Electronic Engineering, 2021, 22 (9): 1179-1194.
[18]	曾筠程, 邵敏华, 孙立军, 等. 基于有向图卷积神经网络的交通预测与拥堵管控[J]. 中国公路学报, 2021, 34(12): 239-248. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL202112018.htm ZENG Y C, SHAO M H, SUN L J, et al. Traffic prediction and congestion control based on directed graph convolution neural network[J]. China Journal of Highway and Transport, 2021, 34(12): 239-248. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL202112018.htm
[19]	DANEL T, SPUREK P, TABOR J, et al. Spatial graph convolutional networks[C]. International Conference on Neural Information Processing, Berlin, Germany: Springer, 2020.
[20]	LI L, ZHU H G, WEN L B, et al. An approach of combining convolution neural network and graph convolution network to predict the progression of myopia[J]. Neural Process Lett, 2023, 55(1): 247-257.
[21]	ALI A, ZHU Y M, ZAKARYA M. Exploiting dynamic spatio-temporal graph convolutional neural networks for citywide traffic flows prediction[J]. Neural Networks, 2022, 145(1): 233-247.
[22]	陈喜群, 曹震, 沈楼涛, 等. 融合路段传输模型和深度学习的城市路网短时交通流状态预测[J]. 中国公路学报, 2021, 34(12): 203-216. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL202112015.htm CHEN X Q, CAO Z, SHEN L T, et al. Short-term traffic-state prediction of urban road networks based on the fusion of a link-transmission model and deep learning[J]. China Journal of Highway and Transport, 2021, 34(12): 203-216. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL202112015.htm
[23]	王康. 基于深度学习的城市短时交通流拥堵预测研究[D]. 武汉: 武汉大学, 2019. WANG K. Research on urban shortterm traffic congestion prediction based on deep learning[D]. Wuhan: Wuhan University, 2019. (in Chinese)