基于神经网络的交通方式选择模型

已有的交通拥挤问题研究大多关注于城市交通的宏观层面,而对于微观层面的居民出行个体研究甚少,这主要是因为与出行个体相关的特征指标难以量化,只能做定性分析,引入神经网络模型即能够识别线性指标又能够识别非线性指标的特性。分析了影响居民出行方式选择的相关因素,这些因素包括出行者自身特性、出行者的出行特性、运输系统特性、出行区域特性和目的地区域特性5类,并建立了神经网络居民出行方式选择预测模型。通过实际调查数据的验证,表明本文模型具有很好的实用性,为城市居民出行方式的选择预测提供了新的思路。

The Option Model of Traffic Mode Based on Neural Network

Formerly traffic congestion researches were mostly focused on the macro aspect of city traffic and few trip individual research were conducted because of being hard to quantify the trip feature index related to individual.With the introduced neural network model,linear and nonlinear indexes can be distinguished,related factors affecting resident trip option can be analysed,including trip self property,trip property,transportation systematic property,regional trip property and destination regional property.Neural network resident trip option forecast model is established.The verification of actual investigation data shows that the model has very good practicality and can be utilized for trip option for city residents.