基于EMD-WDD-MK模型的玛纳斯河年径流预测  

Annual Runoff Prediction of Manas River Based on EMD-WDD-MK Model

作　　者：闫国辉 乔长录[1] 陈伏龙[1] YAN Guo-hui;QIAO Chang-lu;CHEN Fu-long(College of Water Conservancy and Architectural Engineering,Shihezi University,Shihezi 832000,Xinjiang,China)

机构地区：[1]石河子大学水利建筑工程学院,新疆石河子832000

出　　处：《中国农村水利水电》2021年第11期83-89,共7页China Rural Water and Hydropower

基　　金：国家自然科学基金项目(51769030,51769029)。

摘　　要：径流预测是进行水资源优化配置的前提,在区域水资源管理中起着非常重要的作用。为了提高干旱区河流年径流预测精度,本文将经验模态分解法(Empirical Mode Decomposition,EMD)和变分模态分解法(Variational Modal Decomposition,VMD)与加权马尔可夫链进行耦合建模,并引入小波降噪(Wavelet Domain Denoising,WDD),建立EMDMK、EMD-WDD-MK和VMD-MK模型。首先将玛河年径流数据进行分解作为多个分量,并将EMD分解得到的高频分量去噪处理,然后对各分量进行预测并重构得到预测值。通过合格率(QR)、平均绝对误差(MAE)、平均相对误差(MAPE)和均方根误差(RMSE)4种指标,对比分析3种模型的预测精度。结果表明:引入小波降噪的EMD-WDD-MK模型比EMDMK和VMD-MK模型预测精度更高,该耦合模型可为干旱区河流规划和调配提供科学依据。Runoff prediction is the premise of optimal allocation of water resources and plays a very important role in regional water resources management.In order to improve the prediction accuracy of annual runoff in arid regions,Empirical Mode Decomposition(EMD)and Variational Modal Decomposition(Variational Modal Decomposition)methods are used in this paper.VMD is coupled with weighted Markov chains,and Wavelet Domain Denoising(WDD)is introduced to build EMD-MK,EMD-WDD-MK and VMD-MK models.Firstly,the annual runoff data of Mahe River is decomposed into several components,and the high-frequency components decomposed by EMD are denoised.Then,each component is predicted and reconstructed to obtain the predicted value.The prediction accuracy of the three models is analyzed through the four indicators of the qualified rate(QR),mean absolute error(MAE),mean relative error(MAPE)and root mean square error(RMSE).The results show that the EMD-WDD-MK model with wavelet denoising has higher prediction accuracy than EMDMK and VMD-MK models.The coupling model can provide a scientific basis for river planning and allocation in arid areas.

关 键 词：经验模态分解法 变分模态分解法 小波降噪 加权马尔可夫链 径流预测 

分 类 号：TV121[水利工程—水文学及水资源]