基于非集计理论的驾驶适宜性度量模型

Metric Model of Driving Suitability Based on Disaggregate Theory

为量化驾驶人的驾驶适宜性,丰富对其的检测理论和方法,应用非集计理论中的多项分对数(MNL)模型构建驾驶适宜性度量模型。模型以驾驶人一定时间内事故发生次数作为选择肢,以个人固有属性、生理心理属性14项指标作为影响因素,并根据200份实际调查数据标定各影响因素参数。另外,选取60份数据验证该模型。结果显示:14项指标参数检验值均小于1.96,各参数统计学意义显著;模型判定系数为0.364 748,表明模型拟合程度较高;且该模型计算值与统计值最大绝对误差仅为3.3%,表明模型精度较高,可用于预测驾驶适宜性。

In order to quantify the driving suitability, enrich its detection theory and methods, a driving suitability measurement model was built based on MNL model of the disaggregate theory. The number of accidents in a period of time were selected as the alternatives, then individual intrinsic attributes, the physiological and psychological attributes of driver totaled 14 as the influencing factors, then parameters of each factor were demarcated based on actual survey of 200 data. Another 60 driver data were selected to testify the model＇ s validity. The results show that 14 parameters＇ test values are less than 1.96, they have high significance levels in statistics, that the coefficient of determination is 0. 364 748, indicating a high degree of model fit, and the maximum absolute error between statistical values and computational val- ues by MNL is 3.3% , the model has high accuracy to predict the driving suitability.