摘要
在对城市高速公路交通流模型深入研究的基础上 ,针对在不同环境以及时变系统中对复杂非线性大系统的控制 ,提出了一种改进的快速 RBF神经网络算法对交通流进行建模 ,克服了传统的数学模型对交通非线性大系统建模时泛化能力差的缺陷 .该算法是采用 APC- 单路径聚类算法确定 RBF神经网络结构参数的一种快速 RBF神经网络算法 ,网络训练速度快 ,效果良好 ,对实现交通流的在线建模与控制有重要意义 .文中进行了计算机仿真研究 。
Urban freeway traffic modeling was researched.Based on the analysis,for the dynamic environ- ment and time- dependentsystem,a modified speediness RBF neural network was proposed to model the ki- netic freeway traffic flow.The method overcomes the traditional mathematical modeling's shortcoming, thatis ,one model validated in one area or one time may be inaccurate in anotherarea and/ or anothertime. This method used an APC- single path clustering algorithm to determine the parameters of RBF neural network,and a new speediness RBF neural network algotithm was given.Applied in the traffic flow model- ing of freeway,this neural network's fastlearning ability is very importantto realize on- line modeling and control of traffic flow.Computer simulations were used to demonstrate the effectiveness of the use of RBF neural network to model traffic flow.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2000年第5期665-668,共4页
Journal of Shanghai Jiaotong University
基金
国家自然科学基金资助项目! ( 696740 2 3 )
关键词
高速公路
交通流
RBF神经网络
建模
freeway traffic flow
traffic flow kinetics
huge nonlinear system
RBF neural network
