龙羊峡电站下闸蓄水水情长期预报的神经网络方法研究  

Long Time Runoff Forecasting with Neural Network Method for Long Yangxia Hydropower Station

作　　者：张继群[1] 魏文秋[1] 

机构地区：[1]武汉水利电力大学研究生部,武汉430072

出　　处：《应用基础与工程科学学报》1995年第2期99-104,共6页Journal of Basic Science and Engineering

摘　　要：误差反向传播算法(BP算法)是训练多层网络的常用算法,但难于确定网络的有效隐节点数,本文提出了一个有效的算法来自动优选隐层节点数,文中用龙羊峡水电站历年10月份的系列实测资料进行了实例研究,取得了初步令人满意的结果。BP algorithm is probably the most successful training procedure for multi-layer neural nets, but it is difficult to determine the optimal number of hidden units. This paper give a new method to choose the number of hidden node. The result of application in Long Yangxia Hydropower Station is satisfactory.

关 键 词：神经网络 BP算法 隐节点数 径流预报 

分 类 号：TV124[水利工程—水文学及水资源]