摘要
RBF网络具有良好的非线性函数逼近能力,且收敛速度快,而灰色GM(,)静态模型对小样本线性数据的预0N测精度高,将两者有机结合起来,提出了一种新的小样本数据预测方法,即灰色RBF(GRBF)静态预测法。同时,为了提高RBF网络的预测精度和运算效率,文中采用ROLS和后向选择法来训练网络。将GRBF静态预测方法应用到小样本时程数据的预测中,实验结果表明,此预测方法快捷简便,精度高,具有良好的实用性。
RBF(Radial Basis Function) network has good ability in approaching nonlinear function,and its convergent speed is rapid.But Grey Model(0,N) can make an accurate precision of small sample linear data.So,a new method on the basis of RBF and GM(0,N) is proposed in the paper.The method can estimate small sample data more accurately and it is called Grey RBF static prediction method,that is GRBF.In addition,the ROLS and the backward selection algorithm are adopted in the paper to improve precision of prediction.Using the method in practice,the result shows that the method is easy,convenient,accurate and good practicality.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第5期62-64,206,共4页
Computer Engineering and Applications
基金
国家"十五"科技攻关项目资助(项目号:2001BA307B01-02-01)