摘要
基于月径流时间序列的非平稳性特点,首先,将小波变换与函数系数自回归(FAR)模型相结合,利用Mal-lat算法中的Daubechies小波变换和多项式样条估计,建立了基于小波变换的水文预报FAR模型。然后,根据长江芜湖站1972年1月-2006年12月的月径流数据,建立了基于小波变换的FAR模型,并对2007年1月-2008年12月的月径流量进行了预测。计算结果表明:相对于单纯的FAR模型,基于小波变换的FAR模型的拟合与预测误差明显减小,基于小波变换的FAR模型优于FAR模型。
Due to the non-stability of monthly runoff time series,firstly,the paper combined FAR model and wavelet transformation,used the Daubechies Wavelet in the Mallat algorithm and the multi-sample estimate,and established FAR hydrologic forcast model based on the wavelet transform.Then,on the basis of the monthly runoff records dating from Jan.1972 to Dec.2006 of Wuhu Station in Yangtze River,established the FAR model and forecasted manthly runoff from Jan.2007 to Dec.2008.Compared with the simple FAR model,the results showed that the residual of fitting and predition of the FAR model based on the wavelet transformation are significantly reduced.The FAR model based on wavelet transfomation superior to the FAR model.
出处
《水资源与水工程学报》
2011年第4期140-143,共4页
Journal of Water Resources and Water Engineering
关键词
月径流数据
FAR模型
小波变换
多项式样条估计
monthly runoff records
FAR model
wavelet transformation
multi-sample estimation