BP神经网络模型在黄河上游产汇流预报中的应用

The Application of BP Neural Network Model in Runoff Forecasting of the Upper Reaches of the Yellow River

针对黄河上游唐乃亥控制断面以上地形条件复杂、流域面积大且控制站点稀少的实际情况,选取该区域主要产流区吉迈至玛曲段,利用BP神经网络模型对研究区的日径流进行模拟。经分析验证,3层BP神经网络模型简单可靠,误差较小,能较好地反映研究区的产汇流特性,可用于该区域的产汇流模拟及相关预报。

In the light of complex topography、large area and scarce control station above Tang Naihai control section of the upper reaches of the Yellow Rive ,the paper selects the main runoff area of Jimai-Maqu section,and uses BP neural network model to model daily runoff of the study area.After a-nalysis and verification, three layer BP neural network model is simple、reliable and small error, can reflect the characteristics of the studied area.So the model can be used for the runoff simulation and forecast in the region.