集合卡尔曼滤波与神经网络融合的洪水预报研究

Study of flood forecasting by ensemble Kalman filtering and neural network

为了提高河道洪水预报精度,研究了集合卡尔曼滤波法与神经网络模型的河道洪水预报技术。利用龙门、白马寺实测洪水资料预报黑石关洪水并进行了对比检验,讨论了集合卡尔曼滤波与神经网络模型预报洪水的融合过程及其特点。试验结果表明,应用集合卡尔曼滤波技术优于神经网络的预报效果,集合卡尔曼滤波技术与神经网络模型融合可有效提高河道洪水预报的精度。

Ensemble Kalman filter and neural network model were used to study river channel flood forecasting to improve its accuracy. The paper compares the flood data of the Longmen and White horse temple stations with that of the Heishiguan station, and discusses the fusion process and characteristics of these two methods, Test results show that ensemble Kalman filter is better than neural network in flood forecasting. Combination of the two methods effectively improves the accuracy of river channel flood forecasting.