摘要
提出了用多层径向基函数网络（ＭＲＢＦＮ）进行水文时间序列预测的神经网络方法．与通常水文时间序列预测的ＡＲ、ＡＲＭＡ、ＡＲＩＭＡ等模型相比，该方法有较强的非线性处理功能；与多层感知器相比，该方法具有训练时间短和全局收敛性的优点．
In this paper, a new approach based on radial baSis function network is probo forhydIDlogical time series prediction. Compared with the common used models such as AR, ARMA,ARIMA. it's superior in dealing with the nonlinearity, and compared with the multilay6rPerceptron, it has the advantage of global convergence and can considerably decrease computationtabs it has been demonstrated by example that the present approach is a USeful 'tool forhydrological modelling and prediction.
出处
《武汉水利电力大学学报》
CAS
CSCD
1996年第6期1-5,共5页
Engineering Journal of Wuhan University
基金
国家自然科学基金
中国博士后基金
关键词
水文预报
径向基函数网络
泛化性能
自动化系统
hydrological prediction
radial basis function network
generalization performance
