基于门控循环网络时空关联的交通事故预测模型

doi:10.13229/j.cnki.jdxbgxb.20231228

摘要/Abstract

摘要：

为提前预测交通事故是否可能发生，提出了基于ViT、门控循环单元（GRU）和MLP-Mixer相结合的交通事故风险预测模型（GST）。通过ViT进行空间和时间上下文关系建模，对预测目标的帧特征进行增强，提高特征的可分辨性；然后，采用GRU提取出时间关联性，再采用GRU和MLP-Mixer相结合的模式对隐藏层帧特征进行增强，建立和优化时空联系模型，并根据相应的特征帧预测单位时间步长的交通事故置信度分数，预测未来事故发生的概率，进而有效区分危险驾驶和事故驾驶的行为，并进行提前预警。最后，在公开数据集DAD和A3D上对本文模型进行验证，结果表明，本文模型识别准确率优于其他先进算法，两个数据集上AP分别达到了59.9%和94.6%，表现出良好的预测性能和泛化能力；在DAD数据集测试中，将本文算法与DSTA算法进行了对比，在AP相近的情况下，本文算法可将事故发生的预测时间提前2.38 s，提升约13%，具有明显的优越性，可为道路危险预警和安全驾驶提供帮助。

关键词: 智能交通, 事故预测, 事件判别器, 门控循环单元, 时空关联性

Abstract:

To predict the possibility of traffic accidents in advance， a traffic accident risk anticipation model gated recurrent unit spatial-temporal transformer（GST） was established based on the combination of vision transformer（ViT）， gated recurrent unit （GRU）， and MLP-Mixer. By modeling spatial-temporal relational learning through ViT， the frame features of predicted targets were enhanced to improve their distinguishability. On this basis， GRU was used to extract temporal relational， and then GRU and MLP-Mixer were combined to enhance the hidden layer frame features， establishing and optimizing spatiotemporal relational， the confidence score of traffic accidents for each time step was predicted based on the corresponding feature frames to predict the probability of future accidents and effectively distinguish between dangerous driving and accident driving behavior. Finally， the proposed model was validated on the public datasets DAD and A3D， and the results showed that the recognition accuracy of the proposed model was superior to other advanced algorithms. The AP on the two datasets reached 59.9% and 94.6%， respectively， demonstrating good predictive performance and generalization ability. In the DAD dataset， the algorithm proposed was compared to the DSTA model. With similar AP， the proposed algorithm can advance the prediction time of accidents by 2.38 seconds， an increase of about 13%. This indicates a significant advantage and provides assistance for road hazard warnings and safe driving.

Key words: intelligent transportation, accident anticipation, event discriminator, gated recurrent unit, spatial-temporal relational

中图分类号:

TP301.6

江晟,王祎笛,谢睿麟,夏淼磊. 基于门控循环网络时空关联的交通事故预测模型[J]. 吉林大学学报(工学版), 2025, 55(3): 954-962.

Sheng JIANG,Yi-di WANG,Rui-lin XIE,Miao-lei XIA. Traffic accident anticipation baed on spatial-temporal relational learning and convolutional gated recurrent network[J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(3): 954-962.

图/表 6

图1

图2

图3

表1

图4

表2

参考文献 21

1	刘嫣然, 孟庆瑜, 郭洪艳, 等.图注意力模式下融合高精地图的周车轨迹预测[J]. 吉林大学学报: 工学版, 2023, 53(3): 792-801.
	Liu Yan-ran, Meng Qing-yu, Guo Hong-yan, et al. Vehicle trajectory prediction combined with high definition map in graph attention mode[J]. Journal of Jilin University (Engineering and Technology Edition), 2023,53(3):792-801.
2	田彦涛, 黄兴, 卢辉遒, 等. 基于注意力与深度交互的周车多模态行为轨迹预测[J]. 吉林大学学报: 工学版, 2023, 53(5): 1474-1480.
	Tian Yan-tao, Huang xing, Lu Hui-qiu, et al. Multi⁃mode behavior trajectory prediction of surrounding vehicle based on attention and depth interaction[J]. Journal of Jilin University (Engineering and Technology Edition), 2023,53(5): 1474-1480.
3	田彦涛, 许富强, 王凯歌, 等.考虑周车信息的自车期望轨迹预测[J]. 吉林大学学报: 工学版, 2023,53(3): 674-681.
	Tian Yan-tao, Xu Fu-qiang, Wang Kai-ge, et al. Expected trajectory prediction of vehicle considering surrounding vehicle information[J]. Journal of Jilin University (Engineering and Technology Edition), 2023, 53(3): 674-681.
4	李志慧, 孙雅倩, 陶鹏飞, 等. 交通事故后的交通运行风险状态等级预测方法[J]. 吉林大学学报: 工学版, 2022, 52(1): 127-135.
	Li Zhi-hui, Sun Ya-qian, Tao Peng-fei, et al. Prediction method of traffic operation risk level after traffic accident[J]. Journal of Jilin University (Engineering and Technology Edition), 2022,52(1):127-135.
5	Karim M M, Li Y, Qin R. Towards explainable artificial intelligence (XAI) for early anticipation of traffic accidents[J/OL].[2023-10-20].
6	Wu X X, Zhao J W, Wang R Q. Anticipating future relations via graph growing for action prediction[J]AAAI Technical Track on Computer Vision III, 2021, 35(4): 2952-2960.
7	Chan F H, Chen Y T, Xiang Y, et al. Anticipating accidents in dashcam videos[C]∥Computer Vision–ACCV 2016: 13th Asian Conference on Computer Vision, Taipei, Taiwan,China, 2017: 136-153.
8	Karim M M, Li Y, Qin R, et al. A dynamic spatial-temporal attention network for early anticipation of traffic accidents[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(7): 9590-9600.
9	Fatima M, Khan M U K, Kyung C M. Global feature aggregation for accident anticipation[C]∥2020 25th International Conference on Pattern Recognition (ICPR),Milan, Italy, 2021: 2809-2816.
10	Kang M, Lee W, Hwang K, et al. Vision transformer for detecting critical situations and extracting functional scenario for automated vehicle safety assessment[J]. Sustainability, 2022, 14(15): No.9680.
11	Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16x16 words: transformers for image recognition at scale[J/OL].[2023-10-20].
12	张玺君, 余光杰, 崔勇, 等.基于聚类算法和图神经网络的短时交通流预测[J]. 吉林大学学报: 工学版, 2024, 54(6): 1593-1600.
	Zhang Xi-jun, Yu Guang-Jie, Cui yong, et al. Short-term traffic flow prediction based on clustering algorithm and Graph neural network[J]. Journal of Jilin University (Engineering and Technology Edition), 2024, 54(6): 1593-1600.
13	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.
14	Bao W, Yu Q, Kong Y. Uncertainty-based traffic accident anticipation with spatio-temporal relational learning[J/OL]. [2023-10-22]. Computer Vision and Pattern Recognition,
15	Fusco F, Pascual D, Staar P. pNLP-Mixer: an efficient all-MLP architecture for language[J/OL].[2023-10-22].
16	Ren S, He K, Girshick R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J/OL].[2023-10-22].
17	Wojke N, Bewley A, Paulus D. Simple online and realtime tracking with a deep association metric[C]∥2017 IEEE international conference on image processing (ICIP), Beijing, China, 2017: 3645-3649.
18	Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in Neural Information Processing Systems, 2017, 30: 5998-6008.
19	张龙, 徐天鹏, 王朝兵, 等.基于卷积门控循环网络的齿轮箱故障诊断[J]. 吉林大学学报: 工学版,2022, 52(2): 368-376.
	Zhang Long, Xu Tian-peng, Wang Chao-bing, et al. Gearbox fault diagnosis baed on convolutional gated recurrent network[J]. Journal of Jilin University (Engineering and Technology Edition), 2022, 52(2):368-376.
20	Yao Y, Xu M, Wang Y, et al. Unsupervised traffic accident detection in first-person videos[C]∥2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Macau, China, 2019: 273-280.
21	Wang T, Chen K, Chen G, et al. GSC: A Graph and Spatio-temporal Continuity Based Framework for Accident Anticipation[J]. IEEE Transactions on Intelligent Vehicles, 2024, 9(1): 2294-2261.

相关文章 15

[1]	李立,鲍宇健,杨文臣,楚庆玲,汪贵平. 路侧多源感知数据集规范化构建方法[J]. 吉林大学学报(工学版), 2025, 55(2): 529-536.
[2]	杨楠,肖军. 序列二次规划算法下城市智能交通运行节能优化控制[J]. 吉林大学学报(工学版), 2024, 54(8): 2223-2228.
[3]	李立,吴晓强,杨文臣,周瑞杰,汪贵平. 基于路侧毫米波雷达的群体车辆目标识别与跟踪[J]. 吉林大学学报(工学版), 2024, 54(7): 2104-2114.
[4]	张玺君,余光杰,崔勇,尚继洋. 基于聚类算法和图神经网络的短时交通流预测[J]. 吉林大学学报(工学版), 2024, 54(6): 1593-1600.
[5]	井佩光,田雨豆,汪少初,李云,苏育挺. 基于动态扩散图卷积的交通流量预测算法[J]. 吉林大学学报(工学版), 2024, 54(6): 1582-1592.
[6]	杨秀建,贾晓寒,张生斌. 考虑汽车队列动态特性的混合交通流特性[J]. 吉林大学学报(工学版), 2024, 54(4): 947-958.
[7]	马潇驰,陆建. 基于基因表达式编程的高架道路事故实时预测[J]. 吉林大学学报(工学版), 2024, 54(3): 719-726.
[8]	田彦涛,黄兴,卢辉遒,王凯歌,许富强. 基于注意力与深度交互的周车多模态行为轨迹预测[J]. 吉林大学学报(工学版), 2023, 53(5): 1474-1480.
[9]	张惠臻,高正凯,李建强,王晨曦,潘玉彪,王成,王靖. 基于循环神经网络的城市轨道交通短时客流预测[J]. 吉林大学学报(工学版), 2023, 53(2): 430-438.
[10]	高海龙,徐一博,侯德藻,王雪松. 基于深度异步残差网络的路网短时交通流预测算法[J]. 吉林大学学报(工学版), 2023, 53(12): 3458-3464.
[11]	高明华,杨璨. 基于改进卷积神经网络的交通目标检测方法[J]. 吉林大学学报(工学版), 2022, 52(6): 1353-1361.
[12]	张龙,徐天鹏,王朝兵,易剑昱,甄灿壮. 基于卷积门控循环网络的齿轮箱故障诊断[J]. 吉林大学学报(工学版), 2022, 52(2): 368-376.
[13]	邝先验,罗会超,钟蕊,欧阳鹏. 基于天牛须小波神经网络的公交到站时间预测[J]. 吉林大学学报(工学版), 2022, 52(1): 110-117.
[14]	张健,吴坤润,杨敏,冉斌. 智能网联环境下交叉口双环自适应控制模型[J]. 吉林大学学报(工学版), 2021, 51(2): 541-548.
[15]	赵宏伟, 刘宇琦, 董立岩, 王玉, 刘陪. 智能交通混合动态路径优化算法[J]. 吉林大学学报(工学版), 2018, 48(4): 1214-1223.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

实验	RNN	FET	MLP-Mixer	AP/%	mTTA/s
1	GRU	√		69.33	1.53
2	GRU		√	68.75	1.44
3	LSTM	√	√	69.85	1.49
4	GRU	√	√	73.12	1.51

数据集	模型	mTTA/s	AP/%
DAD^［7］	DSA^［7］	1.34	48.1
	DSTA^［8］	2.07	59.2
	GST	2.38	59.9
A3D^［19］	DSA^［7］	2.95	92.3
A3D^［19］	GST	2.42	94.6