短期洪水预报的变结构神经网络模型

Variable Structure Neural Network for Short Term Flood Forecasting

针对神经网络洪水预报模型的结构难以确定的问题 ,应用一种在训练过程中可调整隐层神经元数的算法 ,建立了变结构神经网络洪水预报模型。该方法提供了设计面向问题的网络结构的途径 ,在网络结构设计。

In accordance with the determination of structure of ANN for flood forecasting, this paper applied an algorithm that can modify the structure during the process of learning to build a ANN model for short term flood forecasting. This algorithm provides a way to design the structure of problem oriented neural network. The result shows that it is practical in the design of ANN and improves flood forecasting accuracy.