基于小波分析的径流分级组合预报模型

为进一步提高径流预报精度、降低预报的不确定性，利用小波分析法提取径流系列的概貌和细节成分；采用BP网络模型、RBF网络模型、SVM模型分别模拟预报，进行径流分级。根据不同级别的径流，对预报结果予以变权重组合，构建了基于小波分析的径流分级组合预报模型，并对其预报结果作了分析和总结。宜昌站中长期径流预报结果表明，组合预报模型能够较好地提高预报精度。

Classification and combination prediction model of runoff based on wavelet analysis

To further improve the runoff prediction accuracy and reduce prediction uncertainty,the wavelet analysis is used to extract the outlined component and detail component of the runoff series.Three models,back propagation(BP) neural network model,radial basis function(RBF) network model and support vector machine(SVM) model,were selected to simulate the runoff.Furthermore,the runoff was classified into three parts;the forecasting results are weight-variedly re-composited according to level-different runoff respectively.A combination prediction model of different runoff levels based on wavelet analysis is established,and the results are analyzed and summarized.The results show that the combination prediction model can improve prediction accuracy.