基于熵权-逼近理想解排序法-多群体问询的山区农村公路无信号交叉口安全风险评估模型

基于交通事故统计数据，发现山区农村公路无信号交叉口事故总量大、危险性高的特点。目前为止，山区农村无信号交叉口安全风险暂且没有量化、配套的安全风险评估方法，导致无信号交叉口设计时，评估风险依据不足，故存在很大安全隐患。基于上述问题，采用文献查阅和实地勘察的方法，建立了涵盖交通特性、山区农村道路条件、交通安全设施三大板块以及九大特征的山区农村公路无信号交叉口安全风险模型。采用多元化、多渠道、多群体问询的调查问卷方法，综合不同人群关于交叉口安全风险的评价，对特征指标进行赋值与量化。为避免人为主观因素对特征指标权重的影响，本文运用熵权TOPSIS法确定指标权重，使权重的确定摆脱主观干扰从而更加准确科学。其次，采用加权广义马氏距离与欧式距离贴近度进行对比，数据较为贴合，验证了模型的可行性。最终，结合工程实例，依据指标熵权大小，对安全风险较大的无信号交叉口提出改善措施。研究结果表明，山区农村公路无信号交叉口安全风险指标中线形、视距、坡度占据前三位，权重分别为0.123、0.119、0.114，与交通事故实际调查主要致因排序相近，自从执行改善措施以来，暂无发生交通事故，为定量评估山区农村公路无信号交叉口安全风险提供了新思路。

Safety risk Assessment model based on entropy weight TOPSIS- multi-group inquiry for mountainous rural road unsignalized intersection

Based on the traffic accident statistics, it is found that the total number of accidents and high risk of unsignalized intersections on rural roads in mountainous areas are large. So far, there is no quantified and supporting safety risk assessment method for mountainous and rural unsignalized intersection safety risk, leading to the lack of risk assessment basis in the design of unsignalized intersection, so there are great safety risks. Based on the above problems, using the method of literature review and field investigation, the safety risk model of unsignalized intersection of rural highway in mountainous area is established, which covers three sections of traffic characteristics, rural road conditions in mountainous area, traffic safety facilities and nine features. Using the questionnaire method of diversification, multi-channel and multi-group inquiry, the evaluation of intersection safety risk of different groups is integrated, and the characteristic indexes are assigned and quantified. In order to avoid the influence of subjective factors on the weight of feature index, this paper uses entropy weight TOPSIS method to determine the weight of feature index, so that the determination of weight can get rid of subjective interference and become more accurate and scientific. Secondly, the weighted generalized Mahalanobis distance is compared with Euclidean distance, and the data are relatively fit, which verifies the feasibility of the model. Finally, combined with the engineering example, according to the index entropy weight, put forward improvement measures for the safety risk of non-signalized intersection. The results show that in the safety risk indexes of unsignalized intersections on rural roads in mountainous areas, line shape, visual distance and slope occupy the top three, with weights of 0.123, 0.119 and 0.114 respectively, which are similar to the main causes of traffic accident investigation. Since the implementation of improvement measures, no traffic accident has occurred temporarily. It provides a new idea for quantitatively evaluating the safety risk of unsignalized intersections on rural roads in mountainous areas.