BP与RBF网络在一些非线性曲线拟合上性能的比较  

Performance comparison between some nonlinear curves fitting with BP and RBF

作　　者：严晓明[1] 郑之[1] 

机构地区：[1]福建师范大学数学与计算机科学学院,福建福州350108

出　　处：《福建农林大学学报（自然科学版）》2011年第6期653-656,共4页Journal of Fujian Agriculture and Forestry University:Natural Science Edition

基　　金：福建省自然科学基金资助项目(2009J01273)

摘　　要：对BP与RBF人工神经网络在隐层的传递函数、层数及结点数上的区别进行了比较;对BP与RBF 2种神经网络在给定的3个较为复杂的非线性函数上进行Matlab仿真;并分析比较了这2种人工神经网络在指定参数下对不同非线性函数进行拟合时的拟合精度和时间效率;最后给出BP与RBF网络进行非线性曲线拟合的选择方法.The differences of transform function,numbers of layers and nodes on hidden layer of BR and RBF artificial neural network were compared;the simulation in Matlab of BR and RBF neural networks on three given complex nonlinear functions was put forward;the fitting accuracy and time efficiency of these two kinds of artificial neural network to different nonlinear functions under specified parameters were comparatively analysed;the method of selecting BR and RBF network with nonlinear curve fitting was finally given.

关 键 词：BP RBF 非线性函数 拟合 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]