BP神经网络在闽江十里庵流量预测中的应用

Application of BP Neural Network to Flood Flow Prediction of Shili'an Station in the Minjiang River

将反向传播BP神经网络模型应用于闽江上游流域,建立闽江上游流域洪水最速下降动量反向传播BP神经网络模型;选择闽江上游南平市十里庵水文站最大洪峰流量与其支流建溪七里街站、富屯溪洋口站、沙溪沙县站三个水文站的相应洪水流量建立模型。结果表明,模型预测精度符合要求,可作为防汛部门预测洪水的依据之一,并为汛中防汛会商和指挥调度提供依据。

Gradient Descent with Momentum Back propagation(BP) neural network model is established for prediction flood flow in the upper Minjiang River.The method is applied to modeling of annual maximum peak discharge of Shili'an hydrologic station in the upper Minjiang River and the corresponding flood discharge of three hydrologic stations in branch of the upper Minjiang River,which are Qilijie station in Jianxi River,Yangkou station in Futunxi River and Shaxian station in Shaxi River.The results show that the prediction accuracy of the model is accord with the requirements and it can be used as one of the bases of flood forecasting for the flood control department and provide scientific basis for flood control conference and dispatch and command in flood season.