基于神经网络的静态交通发生率预测方法  

Forecast of the Parking Generating Rate Based on Neural Network

作　　者：熊娟 陈峻[2] 张辉[3] 

机构地区：[1]北京四通智能交通系统集成有限公司,北京100081 [2]东南大学交通学院,南京210096 [3]北京清华城市规划设计研究院,北京100084

出　　处：《交通运输工程与信息学报》2008年第3期71-76,90,共7页Journal of Transportation Engineering and Information

基　　金：国家自然科学基金资助项目(编号:50308005):新形势下城市车辆停放关键方法研究;东南大学优秀青年教师教学科研资助项目

摘　　要：静态交通发生率模型是停车需求预测中较为常用的一种方法,但是,关键指标——静态交通发生率一般难以确定,尤其是它的未来年值;而静态交通发生率的大小又受到多种因素的影响,为了实现对其未来年值的预测,本文分析了影响静态交通发生率的主要因素,采用了具有处理"灰色"、复杂问题的有力工具——神经网络BP算法,提出了以国民生产总值、机动车保有量、用地类型、城市人口和城市面积为输入层节点,静态交通发生率为输出节点的BP计算模型;最后,选用四个城市的停车调查数据来进行神经网络训练、预测。训练模拟结果良好,表明该方法能够比较成功地应用于静态交通发生率的预测。The parking generating rate model is a common algorithm for parking demand prediction, but, the parking generating rate, the crucial index, are difficult to be confirmed, especial their values in the coming year. The parking generating rate is affected by many factors, so, in order to predict its value-to-be, the paper analyzed the primary factors which affects the parking generating rate, and use BP neural network which is a tool capable of dealing with ＂gray＂ , intricate problems, and then presented a BP model whose input nodes are GNP, motor vehicle possessing quantity, land types, the population and area of the city, and whose output is parking generating rate. Lastly, the paper selected the parking survey data of the four cities in china for training and prediction, and the stimulant result is satisfactory, which indicates that BP neural network can be applied to predict parking generating rate successfully.

关 键 词：静态交通发生率 停车需求影响因素 神经网络BP算法 

分 类 号：U491[交通运输工程—交通运输规划与管理]