Risk propensity discrimination of construction accidents under coupling of human-management-environment factors

doi:10.16265/j.cnki.issn1003-3033.2023.S1.1485

Abstract

Abstract:

In order to prevent and reduce construction accidents, a machine learning algorithm was used to identify the risk propensity of accidents from the perspectives of humans, management, and external environments. Firstly, according to the three factors of humans, management, and environments, the scales for measuring the overconfidence of construction personnel, organizational management atmosphere, and construction environment risk levels were designed and tested respectively. Secondly, according to the score of the scale, the K-means algorithm was used to classify different types of accident risk propensity, and comprehensive risk factors were proposed to judge the clustering results. Finally, principal component analysis, Borderline-Smote, and LightGBM algorithms were used to establish the risk propensity discrimination model of construction accidents, and the stability of the model was evaluated by the k-fold cross-validation method. The results show that the prediction accuracy rate of the model is 90.83%. Therefore, the model can effectively identify the risk level of construction accidents, and the accident risks can be divided into three types: low-risk propensity, medium-risk propensity, and high-risk propensity.

Key words: construction accidents, risk propensity, overconfidence, organizational management atmosphere, construction environment, LightGBM algorithm

XIONG Jian, PENG Yiming, CHEN Yao, SONG Zhike, CAI Jing. Risk propensity discrimination of construction accidents under coupling of human-management-environment factors[J]. China Safety Science Journal, 2023, 33(S1): 69-75.

Figures/Tables 13

Tab.1

Tab.2

Fig.1

Fig.2

Fig.3

Fig.4

Tab.3

Fig.5

Tab.4

Tab.5

Fig.6

Fig.7

Tab.6

References 21

[1]	国家统计局. 中国统计年鉴[M]. 北京: 中国统计出版社, 2021: 267-268.
[2]	董利飞, 陈飞宇, 王苗, 等. 建筑施工安全事故死亡人数预测与影响因素分析[J]. 安全与环境学报, 2022, 22(3): 1 430-1 435.
	DONG Lifei, CHEN Feiyu, WANG Miao, et al. Analysis on the forecast of construction death and its influence factors[J]. Journal of Safety and Environment, 2022, 22(3): 1 430-1 435.
[3]	ZHANG Fan, FLEYEH H, WANG Xinru, et al. Construction site accident analysis using text mining and natural language processing techniques[J]. Automation in Construction, 2019, 99: 238-248. doi: 10.1016/j.autcon.2018.12.016
[4]	AYHAN B U, TOKDEMIR O B. Accident analysis for construction safety using latent class clustering and artificial neural networks[J]. Journal of Construction Engineering and Management, 2020, 146(3): DOI: 10.1061/(ASCE)CO.1943-7862.0001762.
[5]	赵挺生, 庞奇志, 姜雯茜. 基于数据库与支持向量机的施工升降机安全风险预测[J]. 中国安全科学学报, 2021, 31(4): 11-17. doi: 10.16265/j.cnki.issn1003-3033.2021.04.002
	ZHAO Tingsheng, PANG Qizhi, JIANG Wenxi. Safety risk prediction of construction elevator based on database and SVM[J]. China Safety Science Journal, 2021, 31(4): 11-17. doi: 10.16265/j.cnki.issn1003-3033.2021.04.002
[6]	冯亚娟, 都思竹, 张竞一. 基于EW-SPA的装配式建筑施工安全评价及预测[J]. 中国安全科学学报, 2019, 29(5): 85-90. doi: 10.16265/j.cnki.issn1003-3033.2019.05.015
	FENG Yajuan, DU Sizhu, ZHANG Jingyi. Safety evaluation and prediction of prefabricated building construction based on EW-SPA[J]. China Safety Science Journal, 2019, 29(5): 85-90. doi: 10.16265/j.cnki.issn1003-3033.2019.05.015
[7]	YANG Jingjing, YE Gui, XIANG Qingting, et al. Insights into the mechanism of construction workers' unsafe behaviors from an individual perspective[J]. Safety Science, 2021, 133: DOI: 10.1016/j.ssci.2020.105004.
[8]	TONG J, FEILER D, IVANTSOVA A. Good choice, bad judgment: How choice under uncertainty generates overoptimism[J]. Psychological Science, 2018, 29(2): 254-265. doi: 10.1177/0956797617731637 pmid: 29283751
[9]	COSTA D F, DE MELO CARVALHO F, DE MELO MOREIRA B C, et al. Bibliometric analysis on the association between behavioral finance and decision making with cognitive biases such as overconfidence, anchoring effect and confirmation bias[J]. Scientometrics, 2017, 111(3): 1 775-1 799. doi: 10.1007/s11192-017-2371-5
[10]	KLUSOWSKI J, SMALL D A, SIMMONS J P. Does choice cause an illusion of control?[J]. Psychological Science, 2021, 32(2): 159-172. doi: 10.1177/0956797620958009 pmid: 33400628
[11]	ALKAISSY M, ARASHPOUR M, ASHURI B, et al. Safety management in construction: 20 years of risk modeling[J]. Safety Science, 2020, 129: DOI: 10.1016/j.ssci.2020.104805.
[12]	ZHANG R P, LINGARD H, OSWALD D. Impact of supervisory safety communication on safety climate and behavior in construction workgroups[J]. Journal of Construction Engineering and Management, 2020, 146(8): DOI: 10.1061/(ASCE)CO.1943-7862.0001881.
[13]	AMIRI M, ARDESHIR A, ZARANDI M H F. Fuzzy probabilistic expert system for occupational hazard assessment in construction[J]. Safety Science, 2017, 93: 16-28. doi: 10.1016/j.ssci.2016.11.008
[14]	XIA Nini, ZOU P X W, LIU Xing, et al. A hybrid BN-HFACS model for predicting safety performance in construction projects[J]. Safety Science, 2018, 101: 332-343. doi: 10.1016/j.ssci.2017.09.025
[15]	CRONBACH L J. Coefficient alpha and the internal structure of tests[J]. Psychometrika, 1951, 16(3): 297-334. doi: 10.1007/BF02310555
[16]	TAHERDOOST H, SAHIBUDDIN S, JALALIYOON N. Exploratory factor analysis; concepts and theory[J]. Advances in Applied and Pure Mathematics, 2022, 27: 375-382.
[17]	钟熙, 孙祥娥. 基于Kmeans++聚类的朴素贝叶斯集成方法研究[J]. 计算机科学, 2019, 46(增1): 439-441.
	ZHONG Xi, SUN Xiang'e. Research on naïve bayes ensemble method based on Kmeans++ clustering[J]. Computer Science, 2019, 46(S1): 439-441.
[18]	刘朝阳, 吕永波, 刘步实, 等. 高速公路桥梁施工过程安全评估及预警研究[J]. 北京交通大学学报, 2019, 43(2): 131-136. doi: 10.11860/j.issn.1673-0291.20180102
	LIU Zhaoyang, LYU Yongbo, LIU Bushi, et al. Study on safety evaluation and early warning of highway bridge construction process[J]. Journal of Beijing Jiaotong University, 2019, 43(2): 131-136. doi: 10.11860/j.issn.1673-0291.20180102
[19]	CHOI J, GU B, CHIN S, et al. Machine learning predictive model based on national data for fatal accidents of construction workers[J]. Automation in Construction, 2020, 110: DOI: 10.1016/j.autcon.2019.102974.
[20]	王芳杰, 王福建, 王雨晨, 等. 基于LightGBM算法的公交行程时间预测[J]. 交通运输系统工程与信息, 2019, 19(2): 116-121.
	WANG Fangjie, WANG Fujian, WANG Yuchen, et al. Bus travel time prediction based on light gradient boosting machine algorithm[J]. Journal of Transportation Systems Engineering, 2019, 19(2): 116-121.
[21]	汪学清, 刘爽, 李秋燕, 等. 基于K折交叉验证的SVM隧道围岩分级判别[J]. 矿冶工程, 2021, 41(6): 126-128.
	WANG Xueqing, LIU Shuang, LI Qiuyan, et al. Classification and discrimination of Surrounding rock of tunnel based on SVM of K-fold cross validation[J]. Mining and Metallurgical Engineering, 2021, 41(6): 126-128.

变量	分类	样本数	占比/%
性别	男	268	94.70
性别	女	15	5.30
年龄/a	18~30	54	19.08
	31~40	79	27.92
	41~50	92	32.51
	≥51	58	20.49
文化水平	小学或以下	98	34.63
	初中	142	50.18
	高中或中专	43	15.19
工龄/a	≤1	40	14.13
	1~3	55	19.43
	3~5	49	17.31
	5~10	70	24.73
	10~20	48	16.96
	≥20	21	7.42
作业区域	高空作业	122	43.11
	地面作业	117	41.34
	隧道作业	44	15.55

维度	Cronbach α系数		KMO统计量
过度乐观	0.831	0.738	0.806^***
认知偏差		0.688
控制幻觉		0.669
安全管理	0.714	0.709	0.801^***
安全交流	0.714	0.664	0.801^***
自然环境	0.710	0.645	0.729^***
作业条件	0.710	0.613	0.729^***

变量	事故风险倾向性	CPO	OMC	CERL
事故风险倾向性	1.000	—	—	—
CPO	-0.856^**	1.000	—	—
OMC	-0.532^**	0.442^**	1.000	—
CERL	-0.158^**	0.183^**	0.079	1.000

超参数	搜索范围	结果
max_depth	(1,7,1)	4
num_leaves	(3,16,1)	7
learning_rate	(0.05,0.10,0.01)	0.07
feature_fraction	(0.6,1.0,0.1)	0.7
min_split_gain	(0.0,1.0,0.1)	0.1
reg_alpha	(0.0,1.0,0.1)	0.1
reg_lambda	(0.0,1.0,0.1)	0

真实情况	预测结果
真实情况	低风险倾向	中等风险倾向	高风险倾向
低风险倾向	39	3	0
中等风险倾向	2	25	3
高风险倾向	0	2	35