摘要
探讨了ANN应用中存在的几个关键问题,提出采用增加监控样本的办法来预防网络过适应现象,以增强网络的概化能力。并以淮河流域为例,将经过改进的BP网络模型应用于流域日径流量预测中,得到了较高的模拟精度。
Artificial neural network (ANN) is an excellent tool processing nonlinear function. A few of key problems at application of ANN model are discussed in this paper. A monitor sample increase method is put forward to prevent overfitting phenomenon, and enhanced expansion ability of the ANN model. A modified BP model is applied to dailyrunoff forecasting for Huai river, and good precision is obtained.
出处
《水电能源科学》
2003年第1期32-34,共3页
Water Resources and Power
基金
河海大学科技创新基金项目(2002)。
关键词
径流量预测
人工神经网络
改进措施
runoff forecasting
artificial neural network
modified measure
