神经网络算法在传统水文模型洪水预报中的应用

Application of Neural Network Algorithms to Flood Forecasting Using Traditional Hydrological Models

为准确地预报洪水以减轻每年洪涝灾害造成的损失。文章提出了一种将BP神经网络算法与半分布式新安江(XAJ)模型相结合的水文模型。以实际案例对提出的算法进行了单周期校正和实时校正的测试。结果表明:改进的水文模型将BP神经网络纳入传统水文模型,可纠正新安江模型的预报误差,提高预测精度、缩短校正的计算时间,具有一定的应用价值。

In order to forecast flood accurately and reduce the loss caused by flood disaster every year.In this paper,a hydrological model that combines the BP neural network algorithm with the semi-distributed Xin'Anjiang(XAJ)model is proposed.The one-cycle correction and real-time correction are tested with a practical case.The results show that the improved hydrological model incorporating BP neural network into the traditional hydrological model can correct the prediction error of the Xin'anjiang model,improve prediction accuracy,and shorten the correction calculation time,which has certain application value.