基于有序选择模型的船舶事故严重度致因分析

doi:10.16265/j.cnki.issn1003-3033.2023.12.0245

摘要/Abstract

摘要：

为探究影响船舶事故严重度的因素,提出一种基于有序选择模型的船舶事故分析方法。首先,基于中国海事局及其直属机构官方网站上发布的604份沿海水域船舶事故调查报告,以船舶事故严重度为因变量,将事故划分为小事故、一般事故、较大以上事故3类,并从船舶、环境、船员、管理和事故5个方面,选取19个事故严重度潜在致因因素。然后,运用有序选择模型(包括有序Logit模型和有序Probit模型)识别显著致因因素。最后,以最佳模型的边际效应量化分析显著致因因素的作用方向和强度。研究结果表明:2个有序模型所识别出的显著致因因素均为船舶作业、强风、未采取安全航速、危险判断不当、导航设备使用不当、配员不足、安全管理不到位、秋季、冬季、沿海航路、碰撞和自沉;所有显著致因因素均满足平行线假设;有序Probit模型的拟合效果优于有序Logit模型;所有显著致因因素均对较大以上船舶事故具有正向效应;就影响程度而言,碰撞最大(边际效应的绝对值为20.44%),危险判断不当最小(4.94%)。

关键词: 船舶事故, 事故严重度, 致因分析, 有序选择模型, 边际效应

Abstract:

In order to explore the factors that affect the severity of ship accidents, a ship accident analysis method based on an ordered choice model was put forward. Firstly, based on 604 investigation reports of ship accidents in coastal waters issued on the official website of the China Maritime Safety Administration and its direct subordinate organizations, as the dependent variable, the severity of ship accidents was divided into minor accidents, general accidents, and larger and above accidents. Secondly, from five aspects of ship, environment, crew, management, and accident, 19 potential influencing factors were selected. Thirdly, the ordered choice model (including the ordered Logit model and ordered Probit model) was utilized to identify the significant factors affecting the severity of ship accidents. In addition, the marginal effect of the best model was used to quantify the influence direction and degree of significant factors on the severity of ship accidents. The findings indicate that the significant influencing factors identified by the two ordered models are the same, including ship operation status, strong wind, failure to adopt safe speed, improper hazard judgment, improper use of navigation equipment, insufficient staffing, inadequate safety management, autumn, winter, coastal routes, collision, and self-sinking. All the significant influencing factors meet the parallel line hypothesis. The fitting effect of the ordered Probit model is better than that of the ordered Logit model. All the significant influencing factors have a positive effect on larger and above ship accidents. In terms of impact degree, collision is the largest (the absolute value of the marginal effect is 20.44%), and improper hazard judgment is the smallest (4.94%).

Key words: ship accidents, accident severity, causation analysis, ordered choice model, marginal effect

江玉杰, 万征, 陈继红. 基于有序选择模型的船舶事故严重度致因分析[J]. 中国安全科学学报, 2023, 33(12): 131-139.

JIANG Yujie, WAN Zheng, CHEN Jihong. Causation analysis of ship accident severity based on ordered choice model[J]. China Safety Science Journal, 2023, 33(12): 131-139.

图/表 5

表1

表2

表3

表4

表5

参考文献 22

[1]	LI Baode, LU Jing, LI Jing. Investigation of accident severity in sea lanes from an emergency response perspective based on data mining technology[J]. Ocean Engineering, 2021, 239: DOI: 10.1016/j.oceaneng.2021.109920.
[2]	LIU Kezhong, YU Qing, YUAN Zhitao, et al. A systematic analysis for maritime accidents causation in Chinese coastal waters using machine learning approaches[J]. Ocean and Coastal Management, 2021, 213: DOI: 10.1016/j.ocecoaman.2021.105859.
[3]	杨利超, 刘敬贤, 刘钊, 等. 价值衰减下水上交通事故致因关联规则挖掘[J]. 中国安全科学学报, 2022, 32(10):127-134. doi: 10.16265/j.cnki.issn1003-3033.2022.10.2087
	YANG Lichao, LIU Jingxian, LIU Zhao, et al. Association rules mining of water traffic accidents causation under value attenuation[J]. China Safety Science Journal, 2022, 32(10):127-134. doi: 10.16265/j.cnki.issn1003-3033.2022.10.2087
[4]	黄常海, 沈佳, 朱冉超, 等. 基于C5.0决策树的船舶交通事故致因分析模型及应用[J]. 中国安全科学学报, 2022, 32(10):90-99. doi: 10.16265/j.cnki.issn1003-3033.2022.10.0483
	HUANG Changhai, SHEN Jia, ZHU Ranchao, et al. Causation analysis model for ship traffic accidents based on C5.0 decision tree and application[J]. China Safety Science Journal, 2022, 32(10):90-99. doi: 10.16265/j.cnki.issn1003-3033.2022.10.0483
[5]	付姗姗, 刘燕平, 席永涛, 等. 北极海冰环境下船舶事故严重程度影响因素分析[J]. 中国安全科学学报, 2019, 29(10): 84-90. doi: 10.16265/j.cnki.issn1003-3033.2019.10.013
	FU Shanshan, LIU Yanping, XI Yongtao, et al. Analysis of influencing factors on severity of ship accidents in ice-covered Arctic waters[J]. China Safety Science Journal, 2019, 29(10):84-90. doi: 10.16265/j.cnki.issn1003-3033.2019.10.013
[6]	王昊, 杨亚东, 马勇, 等. 基于零膨胀负二项回归模型的内河水上交通人命安全分析[J]. 安全与环境学报, 2017, 17(2):408-412.
	WANG Hao, YANG Yadong, MA Yong, et al. Safety analysis over the casualties of maritime transportation and traffic accidents based on the zero-inflated negative binomial regression method[J]. Journal of Safety and Environment, 2017, 17(2): 408-412.
[7]	WENG Jinxian, YANG Dong. Investigation of shipping accident injury severity and mortality[J]. Accident Analysis & Prevention, 2015, 76: 92-101. doi: 10.1016/j.aap.2015.01.002
[8]	WANG Huanxin, LIU Zhengjiang, WANG Xinjian, et al. An analysis of factors affecting the severity of marine accidents[J]. Reliability Engineering & System Safety, 2021, 210: DOI: 10.1016/j.ress.2021.107513.
[9]	WANG Huanxin, LIU Zhengjiang, WANG Xianjian, et al. Analysis of the injury-severity outcomes of maritime accidents using a zero-inflated ordered probit model[J]. Ocean Engineering, 2022, 258. DOI: 10.1016/j.oceaneng.2022.111796.
[10]	蔡玉良, 于淳, 孙旭, 等. 我国沿海船舶事故特征及影响因素分析[J]. 中国航海, 2021, 44(2): 45-52.
	CAI Yuliang, YU Chun, SUN Xu, et al. Characteristics of ship accidents in coastal waters of China and analysis on influencing factors[J]. Navigation of China, 2021, 44(2): 45-52.
[11]	吴琴, 施欣, 陶学宗. 海事事故严重性影响因素及影响程度识别[J]. 交通运输系统工程与信息, 2019, 19(1):185-191.
	WU Qin, SHI Xin, TAO Xuezong. Identifying the factors and their impact levels on severity of maritime traffic accidents[J]. Journal of Transportation Systems Engineering and Information Technology, 2019, 19(1):185-191.
[12]	WANG Likun, YANG Zaili. Bayesian network modelling and analysis of accident severity in waterborne transportation: a case study in China[J]. Reliability Engineering & System Safety, 2018, 180:277-289. doi: 10.1016/j.ress.2018.07.021
[13]	汪飞翔, 杨亚东, 田书冰, 等. 基于SVM的水上交通事故严重程度的影响因素研究[J]. 交通信息与安全, 2018, 36(2):18-23,32.
	WANG Feixiang, YANG Yadong, TIAN Shubing, et al. An analysis of impact factors of accident severity for water transport based on supporting vector machine[J]. Journal of Transport Information and Safety, 2018, 36(2):18-23,32.
[14]	ZHANG Di, YAN Xianping, YANG Zaili, et al. Incorporation of formal safety assessment and Bayesian network in navigational risk estimation of the Yangtze River[J]. Reliability Engineering & System Safety, 2013, 118(10):93-105. doi: 10.1016/j.ress.2013.04.006
[15]	SALIHOGLU E, BEŞIKÇI E B. The use of functional resonance analysis method (FRAM) in a maritime accident: a case study of Prestige[J]. Ocean Engineering, 2021, 219: DOI: 10.1016/j.oceaneng.2020.108223.
[16]	马景峰, 任刚, 李豪杰, 等. 电动自行车与机动车事故严重性影响因素分析[J]. 交通运输系统工程与信息, 2022, 22(2):337-348.
	MA Jingfeng, REN Gang, LI Haojie, et al. Analysis on contributing factors influencing severity of E-bicycle and motor vehicle crashes[J]. Journal of Transportation Systems Engineering and Information Technology, 2022, 22(2):337-348.
[17]	杨文臣, 谢碧珊, 房锐, 等. 山区双车道公路机动车碰撞事故严重度致因比较分析与预测[J]. 交通运输系统工程与信息, 2021, 21(1):190-195.
	YANG Wenchen, XIE Bishan, FANG Rui, et al. Comparative analysis and prediction of motor vehicle crash severity on mountainous two-lane highways[J]. Journal of Transportation Systems Engineering and Information Technology, 2021, 21(1): 190-195.
[18]	CHEN Shiouyu, SHANG Chinshan, LU Kungdon, et al. Enablers of safety citizenship behaviors of seafarers: leader-member exchange, team-member exchange, and safety climate[J]. Maritime Policy & Management, 2023, 50(1):81-96.
[19]	王轶. 海事调查证据的研究[J]. 天津航海, 2017, 38(2):39-41.
	WANG Yi. Research on maritime investigation evidence[J]. Tianjin Navigation, 2017, 38(2):39-41.
[20]	付姗姗, 张悦, 席永涛, 等. 多因素耦合下长江口水域交通事故致因链分析[J]. 中国安全科学学报, 2023, 33(3):60-67. doi: 10.16265/j.cnki.issn1003-3033.2023.03.1367
	FU Shanshan, ZHANG Yue, XI Yongtao, et al. Causal chain of maritime accidents in Yangtze river estuary considering coupling effects of multiple risk factors[J]. China Safety Science Journal, 2023, 33(3):60-67. doi: 10.16265/j.cnki.issn1003-3033.2023.03.1367
[21]	交通运输部. 水上交通事故统计办法[EB/OL]. [2020-10-20]. https://xxgk.mot.gov.cn/2020/jigou/fgs/202006/t20200623_3307634.html.
[22]	LAN He, MA Xiaoxue, QIAO Weiliang, et al. On the causation of seafarers' unsafe acts using grounded theory and association rule[J]. Reliability Engineering and System Safety, 2022, 223: DOI: 10.1016/j.ress.2022.108498.

类别	变量	变量取值与编码	频数
船舶因素	船型	1=干货船,2=液货船,3=渔船,4=其他	548, 83, 190, 76
船舶因素	船舶状态	1=航行,2=停泊,3=作业	696, 80, 121
环境因素	强风	0=否,1=是	764, 133
	能见度不良	0=否,1=是	764, 134
	通航环境复杂	0=否,1=是	247, 650
船员因素	避碰行动不当	0=否,1=是	670, 227
	瞭望不当	0=否,1=是	390, 507
	未采取安全航速	0=否,1=是	673, 224
	危险判断不当	0=否,1=是	501, 396
	导航设备使用不当	0=否,1=是	616, 281
管理因素	船舶证书不全或无效	0=否,1=是	773, 124
	技能经验不足	0=否,1=是	743, 154
	配员不足	0=否,1=是	576, 321
	安全管理不到位	0=否,1=是	727, 170
	安全意识淡薄	0=否,1=是	723, 174
事故因素	事发季节	1=春季,2=夏季,3=秋季,4=冬季	267, 184, 228, 218
	事发时段	0=夜间,1=白天	602, 295
	事发地点	1=沿海航路,2=港口航道,3=停泊水域	483, 246, 168
	事故类型	1=碰撞,2=自沉,3=其他	586, 103, 208

事故类型	虚拟变量1	虚拟变量2	虚拟变量3
干货船	1	0	0
液货船	0	1	0
渔船	0	0	1
其他	0	0	0

类别	变量	容差	方差膨胀因子
船舶	船型	0.846	1.182
船舶	船舶状态	0.676	1.480
环境	强风	0.796	1.256
	能见度不良	0.919	1.088
	通航环境复杂	0.934	1.070
船员	避碰行动不当	0.809	1.237
	瞭望不当	0.481	2.080
	未采取安全航速	0.777	1.288
	危险判断不当	0.881	1.135
	导航设备使用不当	0.678	1.475
管理	船舶证书不全或无效	0.767	1.304
	技能经验不足	0.768	1.302
	配员不足	0.659	1.517
	安全管理不到位	0.828	1.208
	安全意识淡薄	0.894	1.118
事故	事发季节	0.966	1.035
	事发时段	0.921	1.086
	事发地点	0.741	1.350
	事故类型	0.494	2.026

特征因素	变量	变量取值	参考类别	有序Logit模型		有序Probit模型
特征因素	变量	变量取值	参考类别	参数估计	P值	参数估计	P值
船舶	船舶类型	干货船	其他	-0.050	0.836	-0.051	0.720
		液货船		0.042	0.895	-0.009	0.961
		渔船		0.035	0.904	-0.011	0.946
	船舶状态	航行	停泊	0.055	0.835	0.047	0.758
	船舶状态	作业	停泊	0.513	0.097	0.302	0.094
环境	强风	是	否	0.606	0.005	0.348	0.006
	能见度不良	是	否	0.080	0.672	0.057	0.617
	通航环境复杂	是	否	0.121	0.445	0.087	0.343
船员	避碰行动不当	是	否	0.217	0.197	0.138	0.165
	瞭望不当	是	否	-0.182	0.366	-0.111	0.338
	未采取安全航速	是	否	0.374	0.030	0.219	0.030
	危险判断不当	是	否	0.273	0.052	0.160	0.054
	导航设备使用不当	是	否	0.469	0.007	0.278	0.006
管理	船舶证书不全或无效	是	否	-0.044	0.842	-0.038	0.765
	技能经验不足	是	否	-0.127	0.533	-0.091	0.443
	配员不足	是	否	0.484	0.005	0.299	0.003
	安全管理不到位	是	否	0.774	0.000	0.464	0.000
	安全意识淡薄	是	否	0.203	0.251	0.098	0.342
事故	事发季节	夏季	春季	0.081	0.674	0.054	0.636
		秋季		0.439	0.017	0.268	0.013
		冬季		0.429	0.018	0.254	0.017
	事发时段	白天	夜间	0.019	0.897	0.026	0.763
	事发地点	沿海航路	港口航道	0.672	0.000	0.390	0.000
	事发地点	停泊水域	港口航道	0.249	0.300	0.130	0.354
	事故类型	碰撞	其他	1.245	0.000	0.707	0.000
	事故类型	自沉	其他	1.039	0.000	0.621	0.000
分割点1		μ₁		1.032	—	0.589	—
分割点2		μ₂		3.764	—	2.214	—
平行线检验				卡方值=31.09, P值=0.225		卡方值=31.64, P值=0.205
似然比检验				卡方值=177.06, P值=0.000		卡方值=177.83, P值=0.000
拟合优度检验				赤池信息量准则=1 708.338, 贝叶斯信息准则=1 842.712		赤池信息量准则=1 687.136, 贝叶斯信息准则=1 754.323

特征因素	变量	取值	小事故		一般事故		较大以上事故
特征因素	变量	取值	边际值/%	P值	边际值/%	P值	边际值/%	P值
船舶	船舶状态	作业	-7.488	0.020	-0.87	0.110	8.35	0.020
环境	强风	是	-9.598	0.003	-1.11	0.077	10.71	0.003
船员	未采取安全航速	是	-5.994	0.017	-0.69	0.101	6.69	0.017
	危险判断不当	是	-4.428	0.032	-0.51	0.119	4.94	0.032
	导航设备使用不当	是	-5.875	0.012	-0.68	0.089	6.55	0.011
管理	配员不足	是	-6.620	0.003	-0.77	0.070	7.38	0.002
管理	安全管理不到位	是	-12.476	0.000	-1.44	0.044	13.92	0.000
事故	事发季节	秋季	-6.480	0.008	-0.75	0.085	7.23	0.008
	事发季节	冬季	-5.836	0.017	-0.67	0.104	6.51	0.017
	事发地点	沿海航路	-8.906	0.000	-1.03	0.049	9.93	0.000
	事故类型	碰撞	-18.321	0.000	-2.12	0.044	20.44	0.000
	事故类型	自沉	-15.017	0.000	-1.74	0.062	16.75	0.000