摘要
城市公交车辆行程时间预测是公共交通信息服务和运营调度的重要内容,要求较高的实时性和准确性。本文以智能交通运输系统为背景,通过分析公交车辆的行驶特性,基于改进的神经网络模型,建立了公交车辆动态行程时间预测模型,并对比了三种不同输入变量方案的神经网络预测模型,表明该模型具有良好的适用性。此外,将该方法与卡尔曼滤波法的行程时间预测模型进行比较,结果表明,基于神经网络的动态行程时间预测模型精确度较高。
This paper intends to develop the dynamic travel time prediction model for bus vehicles which can be used for bus arrival time prediction. Therefore, real-time transit information service can be provided to passengers and bus operator. After factors that influence bus travel time are analyzed, the dynamic travel time prediction model is set up based on improved Artificial Neural Networks （ANN） , then three scenarios of ANN Prediction model are compared in order to get the better one. Finally, the prediction results are compared with those of Kalman Filter Prediction model. The comparative results reveal that Artificial Neural Networks Prediction model performs better.
出处
《城市公共交通》
2010年第4期37-40,共4页
Urban Public Transport
基金
国家自然科学基金项目(批准号:50208002)
中国博士后科学基金项目(批准号:2002031036)
国家863计划课题(批准号:2008AA11Z202)资助
关键词
公共交通
行程时间预测
神经网络
卡尔曼滤波法
Public Transportation
Travel Time Prediction
Neural Network
Kalman Filter
