基于递归框架的高速公路交通流量实时预测方法

陈宇; 王炜; 华雪东; 赵德

doi:10.3963/j.jssn.1674-4861.2023.01.013

基于递归框架的高速公路交通流量实时预测方法

doi: 10.3963/j.jssn.1674-4861.2023.01.013

陈宇^{1, 2,},
王炜^{1, 2, ,},
华雪东^{1, 2},
赵德^{1, 2}

1.
东南大学交通学院南京 210096
2.
东南大学江苏省现代城市交通技术协同创新中心南京 210096

基金项目:

国家自然科学基金项目 51878166

科技部雄安新区科技创新专项 2022XAGG0126

详细信息

作者简介:
陈宇（1998—），博士研究生. 研究方向：交通运输系统优化. E-mail：230228902@seu.edu.cn

通讯作者:
王炜（1959—），博士，教授. 研究方向：城市虚拟交通系统仿真技术等. E-mail：wangwei_transtar@163.com

中图分类号: U491.1+4
计量
- 文章访问数: 705
- HTML全文浏览量: 248
- PDF下载量: 59
- 被引次数: 0
出版历程
- 收稿日期: 2022-05-22
- 网络出版日期: 2023-05-13

A Recursive Framework-based Approach for Real-time Traffic Flow Forecasting for Highways

CHEN Yu^{1, 2
,},
WANG Wei^{1, 2
, ,},
HUA Xuedong^{1, 2},
ZHAO De^{1, 2}

1.
School of Transportation, Southeast University, Nanjing 210096, China
2.
Collaborative Innovation Center of Modern Urban Traffic Technologies, Southeast University, Nanjing 210096, China

摘要

摘要: 实时与准确的断面交通流量预测是实现高速公路智能化管理与控制的基础。高速公路交通流量预测要求对数据噪声进行有效处理，且需要满足实时性需求。然而，少有研究从实时性的角度对高速公路交通流量预测的准确性进行改善。研究了结合自适应卡尔曼滤波与长短时记忆神经网络（long short-term memory，LSTM）自编码器的高速公路交通流量递归预测框架，可以满足智能交通系统的实时性与准确性需求。收集高速公路的交通流量和速度等历史数据，应用卡尔曼滤波方法进行数据平滑，以提高原始数据的可预测性能；引入无监督机器学习算法LSTM自编码器对交通流量的时变特征进行建模，以提高模型的运算效率；考虑到高速公路交通流量预测的实时性需要，进一步提出递归预测框架，用LSTM自编码器的预测值代替卡尔曼滤波值；根据获取的实时数据，执行自适应卡尔曼滤波算法以修正当前的最佳状态值，并将该修正值输入LSTM自编码器进行迭代预测。选取美国明苏尼达双子城高速公路的实测交通数据进行案例分析，结果表明：所提出的高速公路实时交通流量递归预测框架在计算成本与预测精度2个方面具有相对竞争优势，模型预测的平均绝对百分比误差为5.0%，优于卡尔曼滤波和LSTM自编码器组合模型的7.4%；模型训练时间为85 s，低于标准LSTM模型的101 s。
- 交通运输规划 /
- 高速公路流量 /
- 递归预测 /
- 自适应卡尔曼滤波 /
- LSTM自编码器
Abstract: Real-time and accurate traffic flow forecasting is a prerequisite for intelligent management and control of highways, which requires an effective approach for data processing as well as for meeting the real-time requirement. However, few studies have considered the accuracy of traffic flow forecasting for highways from a real-time perspective. Based on this consideration, a recursive framework for traffic flow forecasting is developed combining adaptive Kalman filter (KF) and long short-term memory (LSTM) autoencoder to meet the real-time and accuracy requirements of intelligent transportation systems. Historical data of traffic flow and speed are adopted, and smoothed by a KF method to enhance the prediction accuracy. An unsupervised machine learning algorithm, LSTM autoencoder, is introduced to model the time-varying characteristics of highway traffic flow efficiently. Considering the real-time requirement of traffic flow forecasting for highways, a recursive forecasting framework is proposed. The output of the KF algorithm is replaced by the predicted value of LSTM autoencoder. Based on the real-time data, the adaptive KF algorithm is conducted to correct the current optimal state value. A case study is conducted based on a real-world traffic dataset collected from the Minnesota Twin Cities, USA. Study results show that the recursive framework of forecasting the highway traffic flow proposed in this study has relatively competitive advantages in terms of both computational cost and prediction accuracy. The mean absolute percentage error of prediction is 5.0% (< 7.4% of the combined KF and LSTM autoencoder model); and total training time is 85 s, which is lower than the standard LSTM (101 s).
- transportation planning /
- highway traffic flow /
- recursive forecasting /
- adaptive Kalman filter /
- LSTM autoencoder

HTML全文

图 1 高速公路路段示意图

Figure 1. Schematic of highway sections

下载: 全尺寸图片幻灯片

图 2 编码器框架示意图

Figure 2. General Architecture of Encoder-Decoder

下载: 全尺寸图片幻灯片

图 3 递归预测框架

Figure 3. Recursive prediction framework

下载: 全尺寸图片幻灯片

图 4 卡尔曼滤波对交通流量和速度的平滑效果图

Figure 4. Smoothing effect on traffic flow and speed

下载: 全尺寸图片幻灯片

图 5 各实验方案在测试集上的预测效果

Figure 5. Performance of the four experimental schemes

下载: 全尺寸图片幻灯片

表 1 各实验方案在测试集上的评价结果

Table 1. Evaluation results of the four experimental schemes

实验方案	预测步长	P_MSE	P_MAPE/%	R²
F1	一步预测	654	9.7	0.96
	二步预测	901	11.3	0.94
	三步预测	1 147	11.9	0.93
	四步预测	1 508	13.0	0.90
	五步预测	1 983	15.2	0.85
F2	一步预测	641	9.5	0.96
	二步预测	909	11.2	0.95
	三步预测	1 087	11.7	0.93
	四步预测	1 410	11.8	0.92
	五步预测	1 998	13.4	0.89
F3	一步预测	544	7.4	0.98
	二步预测	763	9.2	0.97
	三步预测	1 081	9.9	0.96
	四步预测	1 472	10.5	0.94
	五步预测	1 809	11.6	0.91
F4	一步预测	452	5.0	0.99
	二步预测	697	6.9	0.98
	三步预测	912	8.1	0.97
	四步预测	1 290	9.6	0.95
	五步预测	1 564	10.2	0.93

下载: 导出CSV

表 2 标准LSTM与LSTMAE的对比

Table 2. Comparison of LSTM and LSTMAE on F4

实验方案	预测步长	P_MSE/%	R²	训练时间/s*
F4-LSTM AE	一步预测	5.0	0.99	85
	二步预测	6.9	0.98
	三步预测	8.1	0.97
	四步预测	9.6	0.95
	五步预测	10.2	0.93
F4-LSTM	一步预测	6.1	0.97	101
	二步预测	7.2	0.95
	三步预测	9.0	0.95
	四步预测	10.8	0.92
	五步预测	13.5	0.91
F4-SVR	一步预测	8.8	0.95	66
	二步预测	10.3	0.93
	三步预测	11.5	0.92
	四步预测	13.2	0.90
	五步预测	16.7	0.84
F4-ARIMA	一步预测	13.7	0.89	42
	二步预测	15.2	0.85
	三步预测	16.9	0.83
	四步预测	18.0	0.80
	五步预测	20.1	0.75
*注：实验基于GPU（NVIDA GeForce 940MX）与TensorFlow V2.6实现，训练时间包括超参数优化时间。

下载: 导出CSV

参考文献(22)

[1]	VLAHOGIANNI E I, KARLAFTIS M G, GOLIAS J C. Short-term traffic forecasting: where we are and where we're going[J]. Transportation Research Part C: Emerging Technologies, 2014, 43: 3-19. doi: 10.1016/j.trc.2014.01.005
[2]	OH S, BYON Y J, JANG K, et al. Short-term travel-time prediction on highway: a review on model-based approach[J]. KSCE Journal of Civil Engineering, 2018, 22(1): 298-310. doi: 10.1007/s12205-017-0535-8
[3]	DO L N N, TAHERIFAR N, VU H L. Survey of neural network-based models for short-term traffic state prediction[J]. Wiley Interdisciplinary Reviews-Data Mining and Knowledge Discovery, 2019, 9(1): e1285.
[4]	KASHYAP A A, RAVIRAJ S, DEVARAKONDA A, et al. Traffic flow prediction models: a review of deep learning techniques[J]. Cogent Engineering, 2022, 9(1): 2010510. doi: 10.1080/23311916.2021.2010510
[5]	CHAN K Y, DILLON T S, SINGH J, et al. Neural-network-based models for short-term traffic flow forecasting using a hybrid exponential smoothing and Levenberg-Marquardt algorithm[J]. IEEE Transactions on Intelligent Transportation Systems, 2012, 13(2): 644-654. doi: 10.1109/TITS.2011.2174051
[6]	ZHAO L N, WEN X Y, WANG Y P, et al. A novel hybrid model of ARIMA-MCC and CKDE-GARCH for urban short-term traffic flow prediction[J]. IET Intelligent Transport Systems, 2022, 16(2): 206-217. doi: 10.1049/itr2.12138
[7]	GIRAKA O, SELVARAJ V K. Short-term prediction of intersection turning volume using seasonal ARIMA model[J]. Transportation Letters-The International Journal of Transportation Research, 2020, 12(7): 483-490.
[8]	KUMAR B P, HARIHARAN K. Time series traffic flow prediction with hyper-parameter optimized ARIMA models for intelligent transportation system[J]. Journal of Scientific & Industrial Research, 2022, 81(4): 408-15.
[9]	EMAMI A, SARVI M, BAGLOEE S A. Using Kalman filter algorithm for short-term traffic flow prediction in a connected vehicle environment[J]. Journal of Modern Transportation, 2019, 27(3): 222-232. doi: 10.1007/s40534-019-0193-2
[10]	CAI L R, ZHANG Z C, YANG J J, et al. A noise-immune Kalman filter for short-term traffic flow forecasting[J]. Physica A-Statistical Mechanics and its Applications, 2019, 536(C): 122601.
[11]	王科伟, 徐志红. 基于混沌时间序列的道路断面短时交通流预测模型[J]. 交通运输工程与信息学报, 2010, 8(1): 70-74. doi: 10.3969/j.issn.1672-4747.2010.01.014 WANG K W, XU Z H. Chaotic-time-series-based short-term traffic flow forecast model of road cross-section[J]. Journal of Transportation Engineering and Information, 2010, 8(1): 70-74. (in Chinese) doi: 10.3969/j.issn.1672-4747.2010.01.014
[12]	周晓, 唐宇舟, 刘强. 基于卡尔曼滤波的道路平均速度预测模型研究[J]. 浙江工业大学学报, 2020, 48(4): 392-396, 404. https://www.cnki.com.cn/Article/CJFDTOTAL-ZJGD202004006.htm ZHOU X, TAND Y Z, LIU Q. Research on road average speed prediction model based on Kalman filter[J]. Journal of Zhejiang University of Technology, 2020, 48(4): 392-396, 404. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZJGD202004006.htm
[13]	申雷霄, 陆宇航, 郭建华. 卡尔曼滤波短时交通流预测普通国省道适应性研究[J]. 交通信息与安全, 2021, 39(5): 117-127. doi: 10.3963/j.jssn.1674-4861.2021.05.015 SHEN X X, LU Y H, GUO J H. Adaptability of Kalman filter for short-time traffic flow forecasting on national and provincial highways[J]. Journal of Transport Information and Safety, 2021, 39(5): 117-127. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2021.05.015
[14]	KUMAR K, PARIDA M, KATIYAR V K. Short term traffic flow prediction for a non urban highway using artificial neural network[J]. Procedia Social and Behavioral Sciences, 2013, 104: 755-764. doi: 10.1016/j.sbspro.2013.11.170
[15]	CHEN Y Y, CHEN H Y, YE P J, et al. Acting as a decision maker: Traffic-condition-aware ensemble learning for traffic flow prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(4): 3190-3200. doi: 10.1109/TITS.2020.3032758
[16]	许岩岩, 翟希, 孔庆杰, 等. 高速路交通流短时预测方法[J]. 交通运输工程学报, 2013, 13(2): 114-119. XU Y Y, ZHAI X, KONG Q J, et al. Short-term prediction method of freeway traffic flow[J]. Journal of Traffic and Transportation Engineering, 2013, 13(2): 114-119. (in Chinese)
[17]	李巧茹, 赵蓉, 陈亮. 基于SVM与自适应时空数据融合的短时交通流量预测模型[J]. 北京工业大学学报, 2015, 41 (4): 597-602. LI Q R, ZHAO R, CHEN L. Short-term traffic flow forecasting model based on SVM and adaptive spatio-temporal data fusion[J]. Journal of Beijing University of Technology, 2015, 41(4): 597-602. (in Chinese)
[18]	CHEN Y, WANG W, HUA X D, et al. Survey of decomposition-reconstruction-based hybrid approaches for short-term traffic state forecasting[J]. Sensors, 2022, 22(14): 5263.
[19]	LYU Y S, DUAN Y J, KANG W W, et al. Traffic flow prediction with big data: A deep learning approach[J]. IEEE Transactions on Intelligent Transportation Systems, 2015, 16(2): 865-873.
[20]	ZHANG Y N, ZHOU Y H, LU H P, et al. Traffic network flow prediction using parallel training for deep convolutional neural networks on spark cloud[J]. IEEE Transactions on Industrial Informatics, 2020, 16(12): 7369-7380.
[21]	WEI W Y, WU H H, MA H. An AutoEncoder and LSTM-based traffic flow prediction method[J]. Sensors, 2019, 19(13): 2946.
[22]	CHEN Z F, XU J M, LIN Y J, et al. A traffic flow forecasting method regarding traffic network as an digraph[J]. International Journal of Pattern Recognition and Artificial Intelligence, 2021, 35(15): 2159043.