基于Matlab的高速公路交通流RBF神经网络建模  

RBF Neural Network Modeling of Freeway Traffic Flow Based on Matlab

作　　者：王冬丽[1] 吕智林[1] 何小阳[1] 周彦[1] 

机构地区：[1]广西大学,南宁530004

出　　处：《交通与计算机》2005年第5期22-26,共5页Computer and Communications

摘　　要：通过对高速公路宏观动态交通流模型的分析,提出用M atlab神经网络工具箱建立了交通流RBF模型,讨论了RBF网络设计过程中的参数设置,仿真结果表明了RBF模型的有效性。由于M atlab提供的RBF的实现算法具有自适应确定网络结构和无需人为确定网络初始权值的优点,因此减少了网络训练的随机性,提高了训练精度。After analyzing the freeway macroscopic dynamic traffic flow model, this paper established a RBF neural network model based on Matlab neural network toolbox. The problems of data pre-processing and parameters setting were analyzed in detail. The results of simulation proved effectiveness of the RBF model. Because Matlab-based RBF realization method has such advantages as self-adapting network architecture and non-artificially selected initial values, the random of network training was decreased, Thus, the accuracy was naproved.

关 键 词：高速公路交通流模型 MATLAB神经网络工具箱 RBF神经网络 

分 类 号：U491[交通运输工程—交通运输规划与管理]