面向直觉推理的量子效应交通预测算法研究

doi:10.19665/j.issn1001-2400.20240906

摘要/Abstract

摘要：

准确的实时交通预测是实现智能交通系统的核心技术问题。目前已有的预测方法在考虑交通信息的时空特征时,忽略了道路之间空间特征的依赖程度差异,导致预测模型缺乏差异化设计,无法实现对单条道路的精准预测。为了更好地分析道路之间空间特征的依赖程度差异性,设计了面向直觉推理的量子效应交通预测模型。引入直觉推理的思想对路网结构进行编码、组合和比较,分离出在空间特征上高度相关的道路集群,使用量子退火算法优化聚类结果,从而逼近全局最优解。使用华为云研发的MindSpore框架,根据不同的集群构建集群预测模型,专注于每个集群内交通信息的时空特征。在2012年美国洛杉矶高速公路和2021年日本东京1 843条高速公路收集的真实数据集上进行实验,并与历史平均值模型、自回归积分平均移动模型、图卷积网络、门控循环单元和时空图卷积网络进行对比。结果表明,在均方根误差、平均绝对误差、准确率、决定系数和解释差异得分5个指标上均优于上述基线。在两个真实数据集上的均方根误差表现相较基于时空图卷积网络的预测模型分别提升了11.32%和13.86%,为目前交通预测问题提供了一种新的、有效的解决方案。

关键词: 直觉推理, 量子计算机, 量子退火算法, 深度学习, 交通预测

Abstract:

The accurate real-time traffic prediction is the fundamental technological challenge in realizing an intelligent transportation system.Current prediction methods overlook the varying degree of spatial dependence between roads when considering the spatio-temporal characteristics of traffic information,leading to a lack of differentiated design in prediction models and inaccurate predictions for individual roads.To better analyze the differences in spatial features between roads,a quantum effect traffic prediction model is designed for intuitive reasoning.This paper introduces the concept of intuitive reasoning to encode,combine,and compare road network structures,identifying highly correlated road clusters based on spatial features.The quantum annealing algorithm optimizes clustering results towards approximating global optimal solutions.Prediction models are built using the Huawei Cloud's MindSpore framework based on different clusters,focusing on the spatio-temporal characteristics within each cluster.Experiments conducted on real datasets from Los Angeles freeways in 2012 and Tokyo's 1843 freeways in 2021 are compared with various baseline models such as the History Average model,Autoregressive Integrated Moving Average model,Graph Convolutional Network,Gate Recurrent Unit,and Temporal Graph Convolutional Network.The root mean squared error performance on the two real data sets is improved by 11.32%和13.86% compared with the Temporal Graph Convolutional Network,which provides a new and effective solution to the current traffic prediction problem.

Key words: intuitive reasoning, quantum computers, quantum annealing algorithm, deep learning, traffic prediction

中图分类号:

TN915

王潮, 蒋晓锋, 王苏敏. 面向直觉推理的量子效应交通预测算法研究[J]. 西安电子科技大学学报, 2025, 52(1): 152-162.

WANG Chao, JIANG Xiaofeng, WANG Sumin. Research on the quantum effect traffic prediction algorithm oriented towards intuitive reasoning[J]. Journal of Xidian University, 2025, 52(1): 152-162.

图/表 12

图1

图2

图3

图4

图5

表1

表2

表3

图6

图7

图8

表4

参考文献 24

[1]	甘萍, 农丽萍, 张文辉, 等. 一种用于交通预测的注意力时空图神经网络[J]. 西安电子科技大学学报, 2023, 50(1):168-176.
	GAN Ping, NONG Liping, ZHANG Wenhui, et al. Attention Spatial-Temporal Graph Neural Network for Traffic Prediction[J]. Journal of Xidian University, 2023, 50(1):168-176.
[2]	XU Z, YUAN J, YU L, et al. Machine Learning-Based Traffic Flow Prediction and Intelligent Traffic Management[J]. International Journal of Computer Science and Information Technology, 2024, 2(1):18-27.
[3]	YUAN H, LI G. A Survey of Traffic Prediction:from Spatio-Temporal Data to Intelligent Transportation[J]. Data Science and Engineering, 2021, 6(1):63-85.
[4]	ALGHAMDI T, ELGAZZAR K, BAYOUMI M, et al. Forecasting Traffic Congestion Using ARIMA Modeling[C]// 2019 15th International Wireless Communications & Mobile Computing Conference(IWCMC).Piscataway:IEEE, 2019:1227-1232.
[5]	CHANG S Y, WU H C, KAO Y C. Tensor Extended Kalman Filter and its Application to Traffic Prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(12):13813-13829.
[6]	MENG M, TOAN T D, WONG Y D, et al. Short-Term Travel-Time Prediction Using Support Vector Machine and Nearest Neighbor Method[J]. Transportation Research Record, 2022, 2676(6):353-365.
[7]	XU D, WANG Y, PENG P, et al. Real-Time Road Traffic State Prediction Based on Kernel-KNN[J]. Transportmetrica A:Transport Science, 2020, 16(1):104-118.
[8]	LIN G, LIN A, GU D. Using Support Vector Regression and K-Nearest Neighbors for Short-Term Traffic Flow Prediction Based on Maximal Information Coefficient[J]. Information Sciences, 2022, 608:517-531.
[9]	HOCHREITER S, SCHMIDHUBER J. Long Short-Term Memory[J]. Neural Computation, 1997, 9(8):1735-1780. doi: 10.1162/neco.1997.9.8.1735 pmid: 9377276
[10]	YU Y, SI X, HU C, et al. A Review of Recurrent Neural Networks:LSTM Cells and Network Architectures[J]. Neural Computation, 2019, 31(7):1235-1270.
[11]	DEY R, SALEM F M. Gate-Variants of Gated Recurrent Unit(GRU) Neural Networks[C]// 2017 IEEE 60th International Midwest Symposium on Circuits and Systems(MWSCAS).Piscataway:IEEE, 2017:1597-1600.
[12]	LI Z, LIU F, YANG W, et al. A Survey of Convolutional Neural Networks:Analysis,Applications,and Prospects[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 33(12):6999-7019.
[13]	KIPF T N, WELLING M. Semi-Supervised Classification with Graph Convolutional Networks(2016)[J/OL].[2017-02-22]. https://arxiv.org/abs/1609.02907.
[14]	CAO M, LI V O K, CHAN V W S. A CNN-LSTM Model for Traffic Speed Prediction[C]// 2020 IEEE 91st Vehicular Technology Conference(VTC2020-Spring).Piscataway:IEEE, 2020:1-5.
[15]	ZHAO L, SONG Y, ZHANG C, et al. T-GCN:A Temporal Graph Convolutional Network for Traffic Prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 21(9):3848-3858.
[16]	XU C, ZHANG A, XU C, et al. Traffic Speed Prediction:Spatiotemporal Convolution Network Based on Long-Term,Short-Term and Spatial Features[J]. Applied Intelligence, 2022, 52(2):2224-2242.
[17]	ZHU J, WANG Q, TAO C, et al. AST-GCN:Attribute-Augmented Spatiotemporal Graph Convolutional Network for Traffic Forecasting[J]. IEEE Access, 2021, 9:35973-35983.
[18]	BOROS E, HAMMER P L, TAVARES G. Local Search Heuristics for Quadratic Unconstrained Binary Optimization(QUBO)[J]. Journal of Heuristics, 2007, 13:99-132.
[19]	COHEN-ADDAD V, KARTHIK C S. Inapproximability of Clustering in LP Metrics[C]// 2019 IEEE 60th Annual Symposium on Foundations of Computer Science(FOCS).Piscataway:IEEE, 2019:519-539.
[20]	CHAKRABARTI B K, LESCHKE H, RAY P, et al. Quantum Annealing and Computation:Challenges and Perspectives[J]. Philosophical Transactions of the Royal Society A, 2023, 381(2241):20210419.
[21]	WANG S, PEI Z, WANG C, et al. Shaping the Future of the Application of Quantum Computing in Intelligent Transportation System[J]. Intelligent and Converged Networks, 2021, 2(4):259-276.
[22]	WANG C, JI T, WANG S. Online Taxi Dispatching Algorithm Based on Quantum Annealing[C]//CAAI International Conference on Artificial Intelligence. Heidelberg:Springer, 2022:337-347.
[23]	WANG C, ZHANG W, WANG S. An Asymptotically Optimal Public Parking Lot Location Algorithm Based on Intuitive Reasoning[J]. Intelligent and Converged Networks, 2022, 3(3):260-270.
[24]	ZHENG N, LIU Z, REN P, et al. Hybrid-Augmented Intelligence:Collaboration and Cognition[J]. Frontiers of Information Technology & Electronic Engineering, 2017, 18(2):153-179.

模型	Los-loop					EXPY-TKY
模型	R_RMSE	R_MAE	R_Accuracy	R²	R_Var	R_RMSE	R_MAE	R_Accuracy	R²	R_Var
QCT-GCN-C2	4.405 7	2.798 2	0.925 1	0.898 1	0.898 1	14.169 2	9.770 1	0.829 5	0.553 7	0.553 7
QCT-GCN-C3	4.245 9	2.679 5	0.927 8	0.905 3	0.905 4	12.602 9	8.608 7	0.848 4	0.646 9	0.646 9
QCT-GCN-C4	4.269 3	2.692 1	0.927 4	0.904 2	0.904 2	12.107 9	8.251 7	0.854 3	0.674 7	0.675 9
QCT-GCN-C5	4.206 8	2.661 2	0.928 4	0.907 0	0.907 0	11.704 5	7.981 8	0.859 2	0.695 5	0.695 6
QCT-GCN-C6	4.461 3	2.811 8	0.924 1	0.895 4	0.895 5	11.308 3	7.801 6	0.863 9	0.715 8	0.715 9
QCT-GCN-C7	4.283 2	2.677 2	0.927 1	0.903 6	0.903 6	11.014 3	7.545 9	0.867 5	0.730 3	0.730 4
QCT-GCN-C8	4.268 2	2.721 0	0.927 4	0.904 3	0.904 3	10.782 2	7.405 7	0.870 3	0.741 6	0.741 6
QCT-GCN-C9	4.237 4	2.660 2	0.927 9	0.905 7	0.905 7	10.925 4	7.481 1	0.868 6	0.734 6	0.734 7
QCT-GCN-C10	4.346 5	2.725 6	0.926 0	0.900 7	0.900 7	10.594 5	7.276 9	0.872 5	0.750 5	0.750 6

时间/min	基线	Los-loop
时间/min	基线	HA	ARIMA	GRU	GCN	T-GCN	QCT-GCN-C5
5	R_RMSE	7.306 7	10.078 0	5.739 1	7.792 2	4.743 6	4.206 8
	R_MAE	3.878 1	7.701 2	3.272 5	5.352 5	3.116 4	2.661 2
	R_Accuracy	0.875 6	0.827 2	0.902 2	0.867 3	0.919 3	0.928 4
	R²	0.722 4		0.828 9	0.684 3	0.881 8	0.907 0
	R_Var	0.722 5		0.829 6	0.684 4	0.881 7	0.907 0
10	R_RMSE	7.306 7	10.079 2	5.753 7	8.287 8	5.026 6	4.498 8
	R_MAE	3.878 1	7.701 4	3.277 1	5.935 8	3.172 4	2.919 6
	R_Accuracy	0.875 6	0.827 2	0.902 0	0.858 9	0.914 5	0.923 4
	R²	0.722 4		0.828 0	0.643 0	0.868 1	0.894 7
	R_Var	0.722 5		0.828 7	0.643 1	0.868 0	0.895 1
15	R_RMSE	7.306 7	10.080 2	5.754 9	8.590 4	5.543 4	5.225 0
	R_MAE	3.878 1	7.702 9	3.060 2	6.055 4	3.514 1	3.142 3
	R_Accuracy	0.875 6	0.827 1	0.902 0	0.853 7	0.905 6	0.911 0
	R²	0.722 4		0.857 6	0.616 6	0.840 4	0.858 3
	R_Var	0.722 5		0.828 5	0.617 0	0.840 8	0.858 5

时间/min	基线	EXPY-TKY
时间/min	基线	HA	ARIMA	GRU	GCN	T-GCN	QCT-GCN-C10
10	R_RMSE	14.340 7	15.896 2	12.376 2	18.526 8	12.352 6	10.594 5
	R_MAE	9.059 3	11.601 3	8.562 7	13.051 3	8.526 7	7.276 9
	R_Accuracy	0.827 4	0.807 5	0.851 1	0.777 1	0.851 3	0.872 5
	R²	0.544 4		0.662 9	0.239 8	0.663 8	0.750 5
	R_Var	0.544 4		0.667 2	0.239 8	0.667 5	0.750 6
20	R_RMSE	14.340 7	15.896 0	12.742 3	18.526 1	12.388 5	11.199 7
	R_MAE	9.059 3	11.600 9	8.933 4	13.064 3	8.582 0	7.597 2
	R_Accuracy	0.827 4	0.807 5	0.846 7	0.777 1	0.850 9	0.865 2
	R²	0.544 4		0.640 7	0.239 7	0.662 4	0.721 6
	R_Var	0.544 4		0.641 4	0.239 8	0.667 1	0.721 6
30	R_RMSE	14.340 7	15.895 9	14.001 0	18.526 3	13.029 4	11.763 4
	R_MAE	9.059 3	11.600 5	9.446 2	13.064 7	9.160 3	7.950 3
	R_Accuracy	0.827 4	0.807 5	0.831 5	0.777 1	0.843 2	0.858 5
	R²	0.544 4		0.566 2	0.239 7	0.623 9	0.693 7
	R_Var	0.544 4		0.566 9	0.239 8	0.624 0	0.694 2

基线	参数量	Los-loop	EXPY-TKY
基线	参数量	训练时间/(s·epoch^-1)	训练时间/(s·epoch^-1)
GRU	12 672	2.858 2	7.709 8
GCN	768	2.223 3	3.175 6
T-GCN	12 672	4.205 4	12.536 8
QCT-GCN	64 335	4.500 4	13.784 2