数据不平衡的MobileViT网络交通事故严重程度预测模型

doi:10.13229/j.cnki.jdxbgxb.20230614

Abstract

Abstract:

In order to solve the problem of low accuracy of accidents severity prediction caused by data imbalance， a traffic accidents severity prediction model based on deep learning technology was proposed. Machine learning algorithm was used to determine the key variables affecting the severity of accidents， Numerical accidents variables were converted into image data and applied to MobileViT network that combined the convolutional neural network and the self-attention mechanism. Focal loss function was used to adaptively adjust the loss contribution of injury and severe accidents with small data volume， so that the model paid more attention to unbalanced data， and the prediction performance of the model was evaluated by precision， recall and F₁ score. The results show that the overall predictive performance index of the proposed model is above 0.81， which is better than other baseline models， and recall and F1 score are at least increased by 4% and 2.5%. Compared with WGAN-GP-XGBoost and ResNet18 models， MobileViT model has improved recall and F1 score of injury accidents by 25.9% and 4.5% respectively. Compared with the other two models， the prediction performance for severe accidents is the best， with an increase of 8.9%， 4.2%， and 6.7% in precision， recall rate， and F1 score. Compared to other data balancing methods， the MobileViT model enhanced with focal loss as the loss function demonstrates superior performance in imbalanced data prediction.

Key words: engineering of communication and transportation system, accidents severity prediction, imbalanced data, MobileViT, focal loss function

CLC Number:

U491.31

Yi-yong PAN,Xiang-yu XU. Model for predicting severity of accidents based on MobileViT network considering imbalanced data[J].Journal of Jilin University(Engineering and Technology Edition), 2025, 55(3): 947-953.

Figures/Tables 7

Fig.1

Table 1

Fig.2

Table 2

Table 3

Table 4

Fig.3

References 20

1	潘义勇, 吴静婷, 缪炫烨. 老年驾驶员事故严重程度影响因素时间不稳定性分析[J]. 吉林大学学报:工学版,2024, 54(10): 2819-2826.
	Pan Yi-yong, Wu Jing-ting, Miao Xuan-ye. Temporal instability analysis of factors affecting accident severity of elderly drivers[J]. Journal of Jilin University (Engineering and Technology Edition),2024, 54(10): 2819-2826.
2	戢晓峰, 乔新. 建成环境对行人交通事故严重程度的非线性影响[J]. 交通运输系统工程与信息, 2023, 23(1): 314-323.
	Ji Xiao-feng, Qiao Xin. Nonlinear influence of built environment on pedestrian traffic accident severity[J]. Journal of Transportation Systems Engineering and Information Technology, 2023, 23(1): 314-323.
3	Zheng M, Li T, Zhu R, et al. Traffic accident´s severity prediction: a deep-learning approach-based CNN network[J]. IEEE Access, 2019, 7: 39897-39910.
4	吕璞, 柏强, 陈琳. 融合深度反残差与注意力机制的山区高速公路事故严重程度预测模型[J]. 中国公路学报, 2021, 34(6): 205-213.
	Lv Pu, Bai Qiang, Chen Lin. A model for predicting the severity of accidents on mountainous expressways based on deep inverted residuals and attention mechanisms[J]. China Journal of Highway and Transport, 2021, 34(6): 205-213.
5	Wang S, Zhang J, Li J, et al. Traffic accident risk prediction via multi-view multi-task spatio-temporal networks[J]. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(12): 12323-12336.
6	Tian Z, Zhang S. Deep learning method for traffic accident prediction security[J]. Soft Computing, 2022, 26(11): 5363-5375.
7	Mehta S, Rastegari M. Mobilevit: light-weight, general-purpose, and mobile-friendly vision transformer[J/OL].[2023-06-07].
8	Rahim M A, Hassan H M. A deep learning based traffic crash severity prediction framework[J].Accident Analysis & Prevention, 2021,154: No. 106090.
9	Khan M N, Ahmed M M. A novel deep learning approach to predict crash severity in adverse weather on rural mountainous freeway[J]. Journal of Transportation Safety & Security, 2023, 15(8): 795-825.
10	刘文文. 基于深度学习的摩托车事故严重程度预测研究[D]. 重庆: 重庆交通大学交通运输学院, 2022.
	Liu Wen-wen. Research on motorcycle accident severity prediction based on deep learning[D]. Chongqing: College of Traffic & Transportation, Chongqing Jiaotong University 2022.
11	Lin T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]∥Proceedings of the IEEE International Conference on Computer Vision, Venice, Italy, 2017: 2980-2988.
12	Sharma A, Vans E, Shigemizu D, et al. DeepInsight: a methodology to transform a non-image data to an image for convolution neural network architecture[J]. Scientific reports, 2019, 9(1): No. 11399.
13	马永杰, 程时升, 马芸婷, 等. 卷积神经网络及其在智能交通系统中的应用综述[J]. 交通运输工程学报, 2021, 21(4): 48-71.
	Ma Yong-jie, Cheng Shi-sheng, Ma Yun-ting, et al. Review of convolutional neural network and its application in intelligent transportation system[J]. Journal of Traffic and Transportation Engineering, 2021, 21(4): 48-71.
14	Yan H, Ma X, Pu Z. Learning dynamic and hierarchical traffic spatiotemporal features with transformer[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(11): 22386-22399.
15	Li Y, Yang Z, Xing L, et al. Crash injury severity prediction considering data imbalance: a wasserstein generative adversarial network with gradient penalty approach[J]. Accident Analysis & Prevention, 2023, 192: No.107271.
16	Niyogisubizo J, Liao L, Zou F, et al. Predicting traffic crash severity using hybrid of balanced bagging classification and light gradient boosting machine[J]. Intelligent Data Analysis, 2023, 27(1): 79-101.
17	Cicek E, Akin M, Uysal F, et al. Comparison of traffic accident injury severity prediction models with explainable machine learning[J]. Transportation Letters, 2023, 15(9): 1043-1054.
18	Zhang Y, Li H, Ren G. Analyzing the injury severity in single-bicycle crashes: an application of the ordered forest with some practical guidance[J]. Accident Analysis & Prevention, 2023, 189: No. 107126.
19	Mohammadpour S I, Khedmati M, Zada M J H. Classification of truck-involved crash severity: dealing with missing, imbalanced, and high dimensional safety data[J]. PLoS one, 2023, 18(3): No.e0281901.
20	Silagyi II D V, Liu D. Prediction of severity of aviation landing accidents using support vector machine models[J]. Accident Analysis & Prevention, 2023, 187: No. 107043.

Related Articles 15

[1]	Xi-zhen ZHOU,He GONG,Dun-dun LI,Yan-jie JI,Jie YAN. Nonlinear model for impact of built environment on curb parking spaces occupancy [J]. Journal of Jilin University(Engineering and Technology Edition), 2024, 54(9): 2520-2530.
[2]	Ya-qin QIN,Zheng-fu QIAN,Ji-ming XIE. Vehicle cooperative obstacle avoidance strategy driven by CLAM model and trajectory data [J]. Journal of Jilin University(Engineering and Technology Edition), 2024, 54(5): 1311-1322.
[3]	Hong-tao LI,Lin-hong WANG,Jun-da LI. Influence of lighting and speed limit on visual search ability at highway intersections [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(8): 2287-2297.
[4]	Wei-tiao WU,Kun ZENG,Wei ZHOU,Peng LI,Wen-zhou JIN. Deep learning method for bus passenger flow prediction based on multi-source data and surrogate-based optimization [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(7): 2001-2015.
[5]	Zhen-liang LIU,Cun-bao ZHAO,Yun-peng WU,Mi-na MA,Long-shuang MA. Life⁃cycle seismic resilience assessment of highway bridge networks using data⁃driven method [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(6): 1695-1701.
[6]	Hong-fei JIA,Ying-jun XU,Li-li YANG,Nan WANG. League member selection and benefit distribution of commercial vehicles multi⁃modal transportation [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(4): 1060-1069.
[7]	Chao SUN,Hao-wei YIN,Wen-yun TANG,Zhao-ming CHU. Sensor deployment strategy and expansion inference of mobile phone data for traffic demand estimation [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(4): 1070-1077.
[8]	Zhi-wei LIU,Zheng-yun SONG,Jian-rong LIU. Impact of shared autonomous vehicles on choice of subway station connection methods [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(12): 3424-3431.
[9]	Jin XU,Yan-peng WANG,Hai-yuan CHEN,Xiao-bo ZHANG,Cun-shu PAN. Longitudinal driving characteristics and operating speed prediction model of cars on hairpin curves of mountainous roads [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(12): 3432-3445.
[10]	Wen-jing WU,Kang-bei XIONG,Li-li YANG,Si-xu PU. Optimization of bus transfer preferential strategy in passenger corridor based on mental account theory [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(11): 3113-3121.
[11]	Xiu-feng CHEN,Yu-tong GUO,Yue-chen WU,Da-yi QU,Meng-yuan GAO. Multi-objective optimization of trafﬁc signal timings based on dandelion algorithm [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(11): 3122-3129.
[12]	Lang SONG,Jian WANG,Bin-yu YANG,Yong ZHU. Signal timing optimization model for left-turn intersection using double-exit lanes [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(10): 2826-2838.
[13]	Hui-qiu LU,Feng ZHAO,Bo XIE,Yan-tao TIAN. Driver's lane change intention recognition in snow and ice environment based on neural network [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(1): 273-284.
[14]	Wen-bo HUANG,Yan-yan CHEN,Shu-shan CHAI. Decision⁃making behavior of pedestrians in cold competition area under the intervention of mobile information [J]. Journal of Jilin University(Engineering and Technology Edition), 2023, 53(1): 132-140.
[15]	Xian-min SONG,Shu-tian YANG,Ming-xin LIU,Zhi-hui LI. Fluctuation characteristics and prediction method of bus travel time between stations [J]. Journal of Jilin University(Engineering and Technology Edition), 2022, 52(8): 1792-1799.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

Comments

Recommended 0

No Suggested Reading articles found!

排序	变量	重要性	排序	变量	重要性
1	安全气囊	6.234 9	22	天气情况	0.861 5
2	纵向座椅位置	1.660 0	23	性别	0.858 2
3	碰撞类型	1.394 3	24	视野遮蔽	0.837 7
4	驾驶员状况	1.381 8	25	控制方式	0.824 8
5	事故车辆数	1.338 9	26	道路等级	0.818 4
6	碰撞点	1.291 8	27	年龄	0.818 3
7	乘客抛出情况	1.218 6	28	车辆运动状态	0.816 8
8	纵断面线形	1.120 7	29	施工作业	0.807 0
9	道路表面状况	1.070 0	30	道路分隔方式	0.802 6
10	预估速度	1.066 0	31	路口标志	0.802 4
11	超速	1.038 9	32	交叉口类型	0.779 3
12	路肩类型	1.010 7	33	交叉口方向	0.765 9
13	药物使用	0.969 2	34	车道数	0.757 0
14	事故行为	0.966 8	35	行驶方向	0.744 3
15	分心状况	0.950 6	36	路面状况	0.680 9
16	光照情况	0.945 7	37	距离交叉口英里数	0.666 2
17	道路线形	0.920 4	38	肇事逃逸	0.640 6
18	城乡道路	0.889 3	39	醉酒驾驶	0.638 7
19	距离交叉口英尺数	0.882 7	40	环境状况	0.595 4
20	限速	0.868 5	41	事故发生位置	0.557 2
21	事故人员数	0.865 0	42	横向座椅位置	0.000 0

模型	精确率	召回率	F₁分数
MobileViT	0.814 3	0.828 6	0.819 8
WGAN-GP-XGBoost	0.809 7	0.796 4	0.799 8
ResNet18	0.774 6	0.785 8	0.776 8
BBC-LGBM	0.610 3	0.783 9	0.656 0
MLP	0.681 7	0.781 1	0.720 4
ORF	0.413 4	0.638 0	0.459 3
GBDT	0.673 5	0.764 2	0.705 4
SVM	0.518 3	0.660 2	0.546 2

模型	实际类别	预测类别
模型	实际类别	仅财产损失	轻伤	重伤
MobileViT	仅财产损失	749	165	1
	轻伤	106	258	0
	重伤	0	1	23
WGAN-GP-XGBoost	仅财产损失	832	83	0
	轻伤	156	205	3
	重伤	0	2	22
ResNet18	仅财产损失	816	97	2
	轻伤	162	198	3
	重伤	1	1	23

模型	类别	精确率	召回率	F₁分数
MobileViT	仅财产损失	0.876 0	0.818 6	0.846 3
	轻伤	0.608 5	0.708 8	0.654 8
	重伤	0.958 3	0.958 3	0.958 3
WGAN-GP-XGBoost	仅财产损失	0.842 1	0.909 3	0.874 4
	轻伤	0.706 9	0.563 2	0.626 9
	重伤	0.880 0	0.916 7	0.898 0
ResNet18	仅财产损失	0.833 5	0.891 8	0.861 7
	轻伤	0.668 9	0.545 5	0.600 9
	重伤	0.821 4	0.920 0	0.867 9

Model for predicting severity of accidents based on MobileViT network considering imbalanced data

RICH HTML

PDF (PC)