基于Elman神经网络的高速公路限速控制  

Control for Freeway Speed Limitation Based on Elman Neural Network

作　　者：李政[1] 梁新荣[1] 

机构地区：[1]五邑大学信息学院,广东529020

出　　处：《微计算机信息》2007年第28期27-29,共3页Control & Automation

基　　金：广东省自然科学基金资助项目(06300326)

摘　　要：限速控制是高速公路交通控制的重要措施和手段,为了提高限速控制精度,提出Elman神经网络建模方法。阐述了Elman神经网络的原理,根据高速公路主线上车辆群状态、路面状况、气象条件等信息,建立交通流速度限制Elman神经网络模型,Elman神经网络的输入层、上下文层、隐含层和输出层的节点数目分别选为2、12、12和1,采用Levenberg-Marquardt算法对Elman神经网络进行训练,并与RBF神经网络进行仿真对比。结果表明,Elman神经网络和RBF神经网络的训练误差分别为9.99769×10-9和2.38112×10-4,与RBF神经网络相比较,Elman神经网络自适应能力强、泛化能力好,能准确地建立交通流速度限制模型,具有良好的应用前景。School of Information, Wuyi University, Jiangmen, Guangdong 529020, China; 2 College of Traffic and Communications, South China University of Technology, Guangzhou 510640, China Abstract： The control for speed limitation is of great importance in the freeway traffic control. In order to improve the control accuracy of speed limitation, a modeling method of Elman neural network is put forward. The principle of Elman network is formulated and the Elman network model for speed limitation of traffic flow is built based on such information as the number of vehicles on the freeway, the performance of the road surface, and the weather conditions. The node numbers of the input layer, context layer, hidden layer and output layer of the Elman network are selected as 2, 12, 12 and 1 respectively. Levenberg-Marquardt algorithm is used to train the network, and the simulation is carried out in contrast to the RBF network. Simulation results show that the training errors for the Elman network and the RBF network are 9.99769×100and 2.38112×10^-4 respectively. Compared with the RBF network, the Elman network has stronger adaptation and better generalization ability, and can build speed limitation model more accurately. It is found to be potentially applicable in practice.

关 键 词：ELMAN神经网络 高速公路 速度限制 仿真 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]