基于卡口数据的机动车出行生成模型

Generation model of vehicle trip based on traffic camera data

针对出行生成预测方法需要耗费较多的人力物力调查、数据精细化程度较低的现状,综合考虑土地利用类型、土地利用混合度、可达性等指标,使用海量卡口数据获取机动车出行生成量并建立逐步回归分析模型;通过与灰色预测模型耦合的方法对回归模型中人口、就业岗位等灰色变量进行预测,将预测值代入所得线性回归方程得到相应土地类型目标年机动车出行生成的预测量.以武汉市为案例研究对象,分析了典型土地利用类型上的机动车出行发生和吸引的主要影响因素,并对预测值予以验证.研究结果表明:不同土地利用类型的机动车出行产生和吸引的关键影响因素不尽相同,本模型平均预测精度在0.932,可为交通规划用地布局等提供参考.

In view of the current situation where trip generation forecasting methods require abundant manpower and material resources to investigate and the degree of data refinement is low,this research comprehensively considers land use types,land use mixing degree,accessibility and other indicators,and uses massive traffic camera data to obtain the amount of motor vehicle trip generation,thus establishing a stepwise regression analysis model.The population,employment and other gray variables in the regression model are predicted by coupling with the gray prediction model,and the predicted value is substituted into the resulting linear regression equation to obtain the predicted amount of motor vehicle trip generation for the corresponding land type in the target year.The main factors influencing the occurrence and attraction of motor vehicle trips on typical land use types are studied,and the predicted values are verified taking Wuhan as a case object.The research results show that the key influencing factors of motor vehicle trip generation and attraction vary among different land use types.The average prediction accuracy of the proposed model is 0.932,which can provide reference for transportation planning and land layout.