基于旅客出行特征的运输通道客运分担率预测模型--以昌吉赣通道为例

A Passenger Transport Sharing Rate Prediction Model Based on Passenger Travel Characteristics--A Case Study of Nanchang-Ji’an-Ganzhou Corridor

为研究高铁的修建对运输通道内其余客运方式分担比率的影响,尤其是对公路班线客运的影响,以昌吉赣客运通道为例,在经典的多项Logit(MNL)模型基础上,结合旅客出行SP/RP调查数据及不同的特性变量组合和效用函数形式,确定了一种最优的分担率预测模型,该模型能够充分地解释和拟合调查样本数据。首先,基于旅客出行SP/RP调查数据分析昌吉赣通道内的旅客出行行为和出行特征,并分析各影响因素作为模型特性变量的适用性,以此选取了6个影响旅客出行方式选择的主要因素,这些影响因素相互独立且其标定值因旅行起讫点的不同而不同。其次,采用MNL模型作为基础模型,对比分析了在不同特性变量选取方案和不同效用函数选取形式下4组模型拟合昌吉赣通道旅客出行特征数据的能力。随后,通过模型的t检验值、优度比检验值及极大似然估计值确定了一种最优的分担率预测模型,认为应选取旅客年龄、收入、费用来源、费用时间比、发车间隔作为模型的特性变量,且效用函数形式为线性,该模型能够充分地解释和拟合调查样本数据。最后,运用该模型对昌吉赣通道不同区间内的4种客运方式分担率进行计算。计算结果显示:到2020年昌吉赣高铁建成后,南昌至吉安、吉安至赣州以及南昌至赣州区间的公路班线客流分担率分别为7.65%,7.66%,7.34%,而高铁客流分担率分别为73.74%,73.69%,73.78%。通过分析特性变量的参数估计结果和分担率的计算结果,提出不同客运方式尤其是公路班线客运的运营改进策略。

In order to study the impact of the construction of high-speed railway on the sharing rate of the remaining passenger transport modes in transport corridor, especially the impact on the passenger transport of the highway route, taking the Nanchang-Ji’an-Ganzhou corridor for example, based on the classic multiple Logit(MNL) model, combining with passenger travel SP/RP survey data, different characteristic variable combinations and utility function forms, an optimal sharing rate prediction model, which can fully interpret and fit the survey sample data, is determined. First, based on the passenger travel SP/RP survey data, the passenger travel behaviors and characteristics in the Nanchang-Ji’an-Ganzhou corridor are analyzed. The applicability of each influencing factor as model characteristic variable is analyzed. Based on this, 6 main factors affecting passenger travel mode selection are selected. These influencing factors are independent of each other, and their calibration values vary due to the starting and ending points of travel. Second, using the MNL model as the basic model, the abilities of 4 model groups to fit the passenger travel characteristic data of Nanchang-Ji’an-Ganzhou corridor under different characteristic variable selection schemes and different utility function selection forms are compared and analyzed. Afterwards, based on the model’s t-test value, the goodness ratio test value and the maximum likelihood estimation value, an optimal sharing rate prediction model is determined. It is considered that the passenger’s age, income, cost source, cost-to-time ratio, and departure interval should be selected as the characteristic variables of the model, the utility function form should be linear, and the model can fully interpret and fit the survey sample data. Finally, the model is used to calculate the sharing rates of 4 passenger transport modes in different sections of the Nanchang-Ji’an-Ganzhou corridor. The calculation result shows that after the completion of the Nanchang-Ji’an-Ganzhou high-speed railway in 2020, the passenger flow sharing rates of the highway routes from Nanchang to Ji’an, Ji’an to Ganzhou and Nanchang to Ganzhou are 7.65%, 7.66% and 7.34% respectively, while the high-speed passenger flow sharing rates are 73.74%, 73.69% and 73.78% respectively. By analyzing the parameter estimation result of the characteristic variables and the calculation result of the sharing rates, the operational improvement strategies of different passenger transport modes, especially the highway passenger transport mode, are proposed.