基于CEEMDAN-ARMA模型的年径流量预测研究  

Annual Runoff Prediction Based on CEEMDAN⁃ARMA Model

作　　者：张金萍[1,2,3] 许敏 张鑫[1] 肖宏林[1] ZHANG Jinping;XU Min;ZHANG Xin;XIAO Honglin(School of Water Science and Engineering,Zhengzhou University,Zhengzhou 450001,China;Zhengzhou Key Laboratory of Water Resources and Environment,Zhengzhou 450001,China;Henan Province Key Laboratory of Groundwater Pollution Prevention and Remediation,Zhengzhou 450001,China)

机构地区：[1]郑州大学水利科学与工程学院,河南郑州450001 [2]郑州市水资源与水环境重点实验室,河南郑州450001 [3]河南省地下水污染防治与修复重点实验室,河南郑州450001

出　　处：《人民黄河》2021年第1期35-39,共5页Yellow River

基　　金：国家重点研发计划项目(2018YFC0406501);郑州大学2015年优秀青年教师发展基金项目(1521323002);2018年河南省高校科技创新人才支持计划项目(18HASTIT014);中国水利水电科学研究院流域水循环模拟与调控国家重点实验室开放基金项目(IWHR-SKL-KF201802);河南省高等学校青年骨干教师培养计划项目(2017GGJS006)。

摘　　要：为了更好地预测河川径流,提高年径流的预测精度,以黄河源区唐乃亥水文站1956—2016年的实测年径流量为研究数据,采用完全集合经验模态分解(CEEMDAN)和自回归滑动平均模型(ARMA)相结合的方法,建立CEEMDAN-ARMA组合模型,并将组合模型的预测结果与单一的ARIMA模型的预测结果进行对比。结果表明:组合模型的拟合优度大于单一ARIMA模型的拟合优度;组合模型预测的平均相对误差为3.31%,比单一的ARIMA模型的预测精度提高了4.63%。由此可见,CEEMDAN-ARMA模型预测精度高于单一的ARIMA模型,利用CEEMDAN分解得到的IMF分量序列作为ARIMA模型的输入数据可以提高模型的预测精度。In order to better predict river runoff and improve the accuracy of annual runoff prediction,the measured annual runoff of Tang⁃naihai Hydrologic Station in the source area of the Yellow River from 1956 to 2016 was taken as the research data and a CEEMDAN⁃ARMA model was established by combining Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN)with Auto⁃Regres⁃sive Moving Average Model(ARMA)to simulate and predict the annual runoff sequence in the research area,and compared the predicted re⁃sults of the combined model with the single ARIMA model.The results show that the goodness of fit of composite model is better than that of single ARIMA model;the average relative error of the combined model is 3.31%,which is 4.63%lower than that of the single ARIMA mod⁃el.Therefore,the prediction accuracy of CEEMDAN⁃ARMA model is higher than that of the single ARIMA model.The IMFs component se⁃ries which decomposed by CEEMDAN can be used as the input data of ARMA model to improve the prediction accuracy of the model.

关 键 词：径流预测 CEEMDAN ARMA模型 黄河源区 

分 类 号：TV121[水利工程—水文学及水资源] TV882.1[水利工程—水利水电工程]