A methodology for road collapse risk assessment based on data augmentation

doi:10.16265/j.cnki.issn1003-3033.2025.S1.0035

Abstract

Abstract:

To prevent road collapse accidents and minimize losses, it is essential to conduct scientific and effective risk assessments of urban roads. To address the imbalance in ground collapse datasets, a synthetic minority oversampling technique (SMOTE) and a generative adversarial network (GAN) were integrated to enhance the original positive accident samples. Deep learning models including convolutional neural networks (CNN) and multilayer perceptron (MLP) were employed for training, and the trained models could be used for risk identification of urban roads. Experimental validation was conducted using ground collapse incident data from Foshan City (2014-2018). The results demonstrate that data augmentation can significantly improve the model's ability to recognize minority samples (with an average recall rate increase of up to 20%) and solve the overfitting problem when training models with imbalanced data.

Key words: data augmentation, road collapse, risk assessment, deep learning, neural network

CLC Number:

X951交通运输安全

WANG Yizhao, BAI Wenfeng, HE Qinglun, WANG Fei, CHEN Long, HE Sen. A methodology for road collapse risk assessment based on data augmentation[J]. China Safety Science Journal, 2025, 35(S1): 234-238.

Figures/Tables 8

Fig.1

Fig.2

Fig.3

Table 1

Fig.4

Table 2

Table 3

Table 4

References 18

[1]	刘乃钰, 谢昭瑞, 张晓田, 等. 我国城市道路塌陷治理领域管理政策现状分析及思考[J]. 工程建设标准化, 2023(7):72-78.
	LIU Naiyu, XIE Zhaorui, ZHANG Xiaotian, et al. Analysis and reflection on the current situation of urban road collapse management in China[J]. Standardization of Engineering Construction, 2023(7): 72-78.
[2]	KULICZKOWSKA E. The interaction between road traffic safety and the condition of sewers laid under roads[J]. Transportation Research Part D: Transport and Environment, 2016, 48: 203-213.
[3]	周泽林, 张凯, 张恒, 等. 属性识别理论下的岩溶隧道地表塌陷风险评价[J]. 中国安全科学学报, 2022, 32(11):105-112.
	ZHOU Zelin, ZHANG Kai, ZHANG Heng, et al. Risk assessment of surface subsidence in karst tunnels under attribute recognition theory[J]. China Safety Science Journal, 2022, 32 (11): 105-112.
[4]	陈菊艳, 朱斌, 彭三曦, 等. 基于AHP和GIS的矿区岩溶塌陷易发性评估:以贵州林歹岩溶矿区为例[J]. 自然灾害学报, 2021, 30(5):226-236.
	CHEN Juyan, ZHU Bin, PENG Sanxi, et al. Evaluation of the susceptibility of karst collapse in mining areas based on AHP and GIS: a case study of Linxia karst mining area in Guizhou[J]. Journal of Natural Disasters, 2021, 30 (5): 226-236.
[5]	管佳林, 罗周全, 杨彪, 等. 矿区岩溶地表塌陷神经网络预测模型研究[J]. 中国安全科学学报, 2011, 21(9):28-33.
	GUAN Jialin, LUO Zhouquan, YANG Biao, et al. Neural network prediction model for mining area karst surface subsidence[J]. China Safety Science Journal, 2011, 21 (9): 28-33.
[6]	JIANG Juncai, WANG Fei, WANG Yizhao, et al. An urban road risk assessment framework based on convolutional neural networks[J]. International Journal of Disaster Risk Science, 2023, 14(3): 475-487.
[7]	郭林飞, 柴仕琦, 董静怡, 等. 我国城市路面塌陷事故统计分析[J]. 工程管理学报, 2020, 34(2):49-54.
	GUO Linfei, CHAI Shiqi, DONG Jingyi, et al. A statistical analysis of urban road collapse accidents in China[J]. Journal of Engineering Management, 2020, 34 (2): 49-54.
[8]	CHAWLA N V, BOWYER K W, HALL L O, et al. SMOTE: synthetic minority over-sampling technique[J]. Journal of Artificial Intelligence Research, 2002, 16: 321-357.
[9]	GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[J]. Advances in Neural Information Processing Systems (NIPS 2014), 2014, 27:2 672-2 680.
[10]	TABARESTANI E S, HADIAN S, PHAM Q B, et al. Flood potential mapping by integrating the bivariate statistics, multi-criteria decision-making, and machine learning techniques[J]. Stochastic Environmental Research and Risk Assessment, 2023, 37(4): 1 415-1 430.
[11]	ONUR S, SUHA B, CENK D. Mapping regional forest fire probability using artificial neural network model in a mediterranean forest ecosystem[J]. Geomatics, Natural Hazards and Risk 2016, 7: 1 645-1 658.
[12]	LECUN Y, BOSER B, DENKER J S, et al. Backpropagation applied to handwritten zip code recognition[J]. Neural Computation, 1989, 1(4): 541-551.
[13]	JENA R, PRADHAN B, BEYDOUN G, et al. Earthquake hazard and risk assessment using machine learning approaches at Palu, Indonesia[J]. Science of the Total Environment, 2020, 749:DOI:10.1016/j.scitotenv.2020.141582.
[14]	HAKIM W L, REZAIE F, NUR A S, et al. Convolutional neural network (CNN) with metaheuristic optimization algorithms for landslide susceptibility mapping in Icheon, South Korea[J]. Journal of Environmental Management, 2022, 305:DOI:10.1016/j.jenvman.2021.114367.
[15]	PAPADOPOULOU-VRYNIOTI K, BATHRELLOS G D, SKILODIMOU H D, et al. Karst collapse susceptibility mapping considering peak ground acceleration in a rapidly growing urban area[J]. Engineering Geology, 2013, 158: 77-88.
[16]	付康胜. 隧道施工致道路塌陷风险及后果评估体系研究[J]. 四川建材, 2023, 49(11):230-232.
	FU Kangsheng. Research on the risk and consequence assessment system for road collapse caused by tunnel construction[J]. Sichuan Building Materials, 2023, 49 (11): 230-232.
[17]	程鉴基. 广州市地面塌陷特征与防治对策的研究[J]. 铁道工程学报, 2012, 29(1):1-5.
	CHENG Jianji. Research on ground collapse characteristics of Guangzhou and control countermeasures[J]. Journal of Railway Engineering Society, 2012, 29 (1): 1-5.
[18]	钟世英, 丛波日. 城市地面塌陷灾害成因机理分析及分类[C]. 2016年全国工程地质学术年会论文集, 2016:63-67.
	ZHONG Shiying, CONG Bori. The genetic machanism and classification of urban ground coolapses hazard[C]. Proceedings of the 2016 National Annual Conference on Engineering Geology, 2016:63-67.

一级指标	二级指标	来源
地质	距断裂带的距离	www.mnr.gov.cn/sj/
	断裂带的长度	www.mnr.gov.cn/sj/
	土壤类型	fszrzy.foshan.gov.cn
	岩土性质
	历史地质灾害状况
	距附近地质灾害点距离
管道	管道材料
	管道类型
	距附近管道距离
施工	距附近工地距离
施工	施工规模
环境	地区降雨量	data.cma.cn/
	距附近河流距离	www.webmap.cn/
	易涝点	fszrzy.foshan.gov.cn
	低洼地点	fszrzy.foshan.gov.cn
	距附近地铁线的距离	www.foshan.gov. cn/gzjg/fssgdjtj/
	路网密度	www.foshan.gov. cn/gzjg/fssgdjtj/

数据集	RCD	SMOTE-RCD	GAN-RCD
事故点数量	19	969	969
非事故点数量	1 000	1 000	1 000
失衡率	50	1.04	1.04

模型	MLP	MLP (SMOTE)	MLP (GAN)	CNN	CNN (SMOTE)	CNN (GAN)
TP	8	9	16	11	16	1.00
FP	2	37	672	1	30	0
TN	998	963	328	999	970	1000
FN	11	10	3	8	3	18
准确率	0.99	0.95	0.34	0.99	0.97	0.98
平均召回率	0.71	0.72	0.59	0.79	0.91	0.53
特异性	1.00	0.96	0.33	1.00	0.97	1.00