基于小波去噪和WNN-ARIMA组合模型的年径流预测

为解决单一的小波神经网络预测精度不高的问题,提出一种新的基于小波去噪和WNN-ARIMA组合模型,应用小波阈值去噪法对小波神经网络的输入值进行预处理,同时对模型残差值进行ARIMA模型修正.利用该组合模型对洮河流域下巴沟站年径流量进行预测,预测趋势和预测值与原始实测数据吻合度高,表明此组合模型可靠性强,可以有效预测年径流量,以期为洮河流域和其他流域的年径流量预测提供新方法,为水利工程建设和水资源优化配置提供依据.

Annual Runoff Prediction Based on Wavelet Denoising and WNN-ARIMA Combined Model

In order to solve the problem of low prediction accuracy of a single wavelet neural network,this paper proposed a new combined model based on wavelet denoising and WNNARIMA.The wavelet threshold denoising method is used to preprocess the input value of the wavelet neural network,and the residual values of the model are corrected by the ARIMA model.The combined model is used to predict the annual runoff at the Xiabagou Station in the Taohe River Basin and the predicted trend and values are in good agreement with the original measured datas,which show that the combined model is highly reliable and can effectively predict the annual runoff.This combined model is expected to provide a new method for the annual runoff prediction of the Taohe River Basin and other basins,and provide a basis for the construction of water conservancy projects and the optimal allocation of water resources.