基于因子分析与模糊聚类的出租车行驶安全评估

Taxi driving safety assessment based on factor analysis and fuzzy clustering

为深入分析营运车辆在行驶过程中存在的问题,有针对性地改善其运营水平,增强道路安全主动防御能力,以新疆某地区出租车运行数据为研究对象,提取与运行特征有关的9个驾驶行为参数,采用因子分析法将其表述为3个更为明确的驾驶行为变量,并基于模糊C均值(FCM)聚类对特征指标进行聚类分析;通过CRITIC赋权法与线性回归法构造驾驶行为综合得分模型,实现对驾驶行为预警指数等级的划分。结果表明:基于因子分析法可把车辆运行状态分为超速行驶、变速行驶和急加速行驶3种情况;通过驾驶员综合得分并利用FCM算法将驾驶行为分为无预警状态、轻度预警状态、中度预警状态、重度预警状态4个等级,每个等级的阈值区间为0~0.17、0.17~0.39、0.39~0.79、0.79~1。研究表明,通过因子分析与FCM聚类相结合的方法能根据出租车的运行状态自动识别出危险系数较高的车辆并对其发出安全预警提示,从而提高对道路交通安全的管理。

In order to deeply analyze the problems of operating vehicles in the process of running,operation level should be improved and the active def ense capability of road safety should be enhanced.Taking the running data of taxis in a region of Xinjiang as the research object,nine driving behavior parameters related to the running characteristics were extracted and expressed into three more specific driving behavior variables by factor analysis method,and cluster analysis was carried out on the characteristic indicators based on FCM clustering.By means of CRITIC weighting method and linear regression method,a comprehensive score model of driving behavior is constructed,and the grade of driving behavior warning index is divided.The results show that based on the factor analysis method,the vehicle running state can be divided into overspeed driving,variable speed driving and rapid acceleration driving.By pilot comprehensive scores and using the FCM algorithm is driving behavior can be divided into state without any warning,warning state of mild,moderate state of early warning,severe warning state,each level of the threshold value range is of 0~0.17,0.17~0.39,0.39~0.79,0.79~1,and respectively.It can be seen that the combination of factor analysis and FCM clustering method can automatically identify vehicles with high risk coefficient according to the running state of taxis and issue safety warning to them,thus improving the management of road traffic safety.