基于熵权法和BP神经网络的道路交通安全度评价模型及应用研究  

Research on Construction and Application of Road Traffic Safety Model Based on BP Neural Network

作　　者：康维 龚雪蕾 KANG Wei;GONG Xuelei(Wuhan University of Technology,Wuhan 430070,China;Kunming Medical University,Kunming 650500,China)

机构地区：[1]武汉理工大学土木工程与建筑学院,湖北武汉430070 [2]昆明医科大学公共卫生学院,云南昆明650500

出　　处：《洛阳理工学院学报（自然科学版）》2019年第4期11-17,60,共8页Journal of Luoyang Institute of Science and Technology：Natural Science Edition

基　　金：国家自然科学基金项目(61863019)

摘　　要：为评价全国各省市地区道路交通安全水平,从公路网密度、道路交通事故严重情况、车辆数3个方面,构建道路交通安全度测评体系。从数据可获得性、一致性原则等方面综合考虑,选择了10项指标作为道路交通安全度测评指标,并进一步基于熵权法,对各指标的权重进行了计算。考虑到道路交通系统的复杂性、随机性,基于BP神经网络技术,进一步建立了包含输入层、隐含层及输出层的道路交通安全度评价模型,并选取北京、天津直辖市及10个省市的道路安全度指标数据作为输入训练样本,以此来对各个省市的道路交通安全度进行测评。结果表明:构建的道路交通安全度评价模型得到的评价结果与各省市情况基本吻合,可有效地评估各省市道路交通安全风险。In order to evaluate the level of regional road traffic safety, the evaluation system of road traffic safety degree is constructed from three aspects of road network density, road traffic accident severity and number of vehicles. Considering comprehensively data availability and consistency principle, 10 indexes are selected as the evaluation index of road traffic safety degree. Furthermore, based on the entropy weight method, the weights of each index are carried out. Considering the complexity and randomness of road traffic system, a road traffic safety evaluation model including input layer, hidden layer and output layer and so on are further established based on neural network technology. The road safety index data of 10 provinces such as Beijing and Tianjin are selected as input training samples to evaluate the road traffic safety degree of each province. The results show that the evaluation results obtained by the risk assessment model are basically consistent with the number and severity of accidents in each province, which can be used to evaluate the road traffic safety risk in this region more effectively.

关 键 词：道路交通 安全度评价 BP神经网络 熵权法 

分 类 号：U493[交通运输工程—交通运输规划与管理]