基于变分模态分解和引力搜索算法的径流预测模型研究  

Research on runoff prediction model based on variational mode decomposition and gravity search

作　　者：李晓春 包苑村 罗军刚[2] 左岗岗 LI Xiaochun;BAO Yuancun;LUO Jungang;ZUO Ganggang(River and Reservoir Work Center of Shaanxi Province,Xi’an 710018,China;State Key Laboratory of Eco-hydraulics in Northwest Arid Region of China,Xi’an University of Technology,Xi’an 710048,China)

机构地区：[1]陕西省江河水库工作中心,陕西西安710018 [2]西安理工大学西北旱区生态水利国家重点实验室,陕西西安710048

出　　处：《西安理工大学学报》2021年第2期194-201,共8页Journal of Xi'an University of Technology

基　　金：国家自然科学基金资助项目(51679186);陕西省自然科学基础研究计划资助项目(2019JLZ-15);陕西省水利科技计划资助项目(2018slkj-4);西北旱区生态水利国家重点实验室自主研究课题资助项目(2019KJCXTD-5)。

摘　　要：针对月径流数据的数据量小、高度非线性和非平稳性特点,提出了一种基于变分模态分解(VMD)、引力搜索(GSA)与支持向量回归机(SVR)的组合预测模型。首先对月径流样本进行Mann-Kendall(M-K)检验,找出突变点划分训练集与测试集;然后用VMD分别对训练集与测试集进行分解;最后对训练集数据进行训练,利用GSA对SVR的三个主要参数进行寻优,并对测试集的预测结果进行评价。将该模型应用于渭河流域咸阳站与临潼站的月径流预测中。结果表明,VMD-GSA-SVR模型相较于其他模型,有着更高的预测精度与泛化能力,更加符合预测的实际情况,在渭河流域具有较好的可行性与适应性。For the small data volume,highly non-linear and non-stationary characteristics of monthly runoff data,a combined prediction model based on variational mode decomposition(VMD),gravitational search algorithm(GSA)and support vector regression machine(SVR)is proposed.Firstly,Mann-Kendall(M-K)test is carried out on monthly runoff samples to find out break points and divide training set and test set.Then,VMD is used to decompose training set and test set separately.Finally,training set data is trained and GSA is used to optimize the three main parameters of SVR and to evaluate the prediction results of test set.The model is applied to the monthly runoff forecast of Xianyang Station and Lintong Station in Weihe River Basin.The results show that the VMD-GSA-SVR model has higher prediction accuracy and generalization ability than other ones,is more consistent with the actual situation of prediction,and has better feasibility and adaptability in the Weihe River Basin.

关 键 词：径流预测 优化算法 信号分解 渭河流域 

分 类 号：TV12[水利工程—水文学及水资源]