小波神经网络在径流预测中的应用

将小波分析与传统的BP神经网络模型进行组合,提出了一种新的径流中长期预测方法。该方法对年径流序列进行Mallat小波分解,将分解后得到的不同尺度下的低频成分和高频成分分别进行Mallat算法重构,对重构系列采用BP神经网络模型进行预测。采用黄河三门峡站1470-2002年的年径流资料进行模型的预测和检验,并与传统的BP神经网络模型进行比较,研究结果表明小波神经网络在径流预测中具有较好的预报精度,可以成功地用于径流模拟和预测。

Application of Wavelet Neural Network Model in Runoff Prediction

Through combining the wavelet and the BP neural network model, a new prediction meth-od is developed for medium - long term runoff prediction. To begin with, the annual runoff series is decomposed systematically based on Mallat method. Second, low - frequency components and high frequency components which decomposed of different scales were reconstructed to Mallat algorithm . And finally, the factored series are used to forecast with the BP neural network model. This paper is applied to annual runoff prediction in Sanmenxia station of Yellow River by 1470 to 2002 to predict and test, which is compared with traditional BP neural network. The results show that wavelet neural network model is effective in practice, and also improves the forecasting precision.