运输能力约束条件下的旅客出行选择Logit模型

Logit model for travel mode choice with traffic capacity constraints

为了提升旅客运输通道交通需求预测的准确性,针对现有旅客出行选择模型未充分考虑交通方式运输能力约束的不足,在深入分析旅客出行选择机理和运输能力约束机制的基础上,通过经典多项Logit(MNL)模型效用函数的优化设计,引入惩罚因子表征运输能力对旅客出行选择的约束,建立运输能力约束条件下的旅客出行选择Logit模型(约束MNL模型),并设计模型求解算法预测各交通方式的分担率。以西宝(西安—宝鸡)客运通道为对象进行实例分析,通过2种MNL模型预测结果的对比分析,验证约束MNL模型预测性能的优越性。研究结果表明:在旅客出行选择过程中,交通方式运输能力的约束具有普遍性,是有效交通需求预测不可忽略的重要因素,约束MNL模型考虑了运输能力对旅客出行选择的影响,更符合旅客出行选择的决策过程,从机理上为提升交通需求预测的准确性提供了可靠保障;惩罚因子反映了运输能力约束对旅客出行选择的影响,代表了运输能力约束条件下运输服务质量的下降和旅客出行效用的损失;通过惩罚因子的合理赋值,建立旅客出行选择概率重新分配机制,能有效模拟旅客出行方式的转换、控制旅客出行选择的概率;与传统MNL模型相比,约束MNL模型表现出了更优异的预测性能,能始终将预测结果控制在由运输能力决定的分担率上限范围内,预测结果符合实际、科学有效,能够为旅客运输通道的网络布局优化、运输组织设计等提供可靠数据支持。

In order to improve the accuracy of traffic demand forecasting for passenger transport corridors,addressing the shortcoming that the traffic capacity constraints are not adequately considered in the traditional passengers'mode choice models,the mechanism of passengers'mode choice and constraints from traffic capacity were deeply analyzed,and a penalty factor was introduced through the utility function optimization of the classical MNL model,to characterize the constraints of traffic capacity on passengers'mode choice,a constrained MNL model was proposed,and the corresponding algorithm was designed to forecast the share of each transportation mode.The XiBao(Xi'an to Baoji)passenger corridor was taken as an example to verify the superior performance of the constrained MNL model by comparative analyzing the forecasting results of the two models.The results show that in the process of passengers'mode choice,the constraints of traffic capacity are universal,which is an important factor that cannot be ignored for effective traffic demand forecasting,the constrained MNL model takes into account the impact of traffic capacity on passengers'mode choice,which is more in line with the decisionmaking process of passengers'mode choice,and provides a reliable guarantee for improving the accuracy of traffic demand forecasting from the mechanism.The penalty factor reflects the impact of traffic capacity constraints on passengers'mode choice,and represents the decline in transportation service quality and the loss of passengers'utility.Through the reasonable assignment of the penalty factor,it can redistribute the passengers'choice probability,effectively simulate the shift of passengers'mode choice and control the probability of passengers'mode choice.Compare with the traditional MNL model,the constrained MNL model shows better performance,which can always control the forecasting results within the upper limit of the share determined by the traffic capacity,and the results are realistic,scientific and effective,which can provide reliable data support for the optimization of the network layout and the design of the transportation organization.