摘要
为探究平面交叉口不同严重程度交通事故影响因素的关联特性,从驾驶人、环境、道路、车辆4个维度出发,建立了带约束的二进制粒子群-蚁群算法以挖掘交叉口事故严重程度的关联规则。首先通过二进制粒子群算法确定最优提升度和支持度阈值,再利用蚁群算法挖掘最大频繁项集,并增加规则前项与后项的约束,以提高关联规则挖掘效率。对2022条平面交叉口事故数据的分析结果表明,提出的BPSO-ACA算法可在精准识别潜在因素与事故等级关联结果的基础上减少冗余无效关联规则,并将规则挖掘效率提升17%~29%。驾驶人年龄、分心状态、交叉口形态、交叉口车道数和天气均与交叉口事故严重程度有强关联性;车道数少是导致轻微事故升级为一般事故的重要因素;路面湿滑是导致交叉口事故升级为重大事故的关键因素,尤其在雨雪雾等不良天气条件下,机非事故为重大等级的可能性最高。研究成果可为交通管理部门的主动防控措施提供理论指导,减少交叉口事故数量及降低事故严重程度。
To access the causing factors of accidents severity at intersections,a constraint Binary Particle Swarm Optimization-Ant Colony Algorithm(BPSO-ACA)is established to acquire the association rules among the causing factors in the prospects of driver,environment,road,and vehicle.Firstly,the BPSO algorithm is used to determine two association rule generation thresholds,minimum support,and minimum confidence.Secondly,the ACA algorithm is applied to mine the frequent itemsets.Finally,dimensional constraints on the antecedent and consequent terms of the rule are added to remove invalid rules.The proposed algorithm makes use of heuristic information and the parallel computing features of the ant system to overcome the disadvantages of traditional algorithms,such as low mining efficiency and a large number of redundant rules caused by repeatedly scanning the database when searching frequent itemsets.The results using 2022 intersection accident data from a province in China entries indicate that the proposed BPSO-ACA algorithm can identity correlating potential factors in accident severity accurately.Compared with the improved Apriori algorithm,the proposed BPSO-ACA algorithm can avoid the generation of invalid rules and improve the efficiency of mining rules by 17%-29%,and the efficiency improvement is even better when the support level is low.Driver age,distraction status,intersection geometry,lane number,and weather condition are strongly related to intersection accident severity.Limitation of lane number is a critical factor in the transformations of severity that from slight to ordinary.Wet pavement condition is a key factor in escalating accidents at intersections into severe accidents.Especially in adverse weather conditions such as rain,snow,and fog,the possibility of severe accidents is the highest for motor and non-motor vehicle accidents.The results can provide theoretical guidance to traffic management departments for active prevention,which can reduce the accident number and severity at intersections.
作者
徐金华
李岩
张玉婷
XU Jin-hua;LI Yan;ZHANG Yu-ting(College of Transportation Engineering,Chang’an University,Xi'an710064,China)
出处
《安全与环境学报》
CAS
CSCD
北大核心
2022年第3期1412-1420,共9页
Journal of Safety and Environment
基金
国家自然科学基金项目(71901036)
国家重点研发计划项目(2017YFC0803906)
陕西省自然科学基金项目(2020JM-222,2020JM-237)。
