基于经验模态分解的年径流组合预测模型

Combination Prediction Model of Annual Runoff Based on Empirical Mode Decomposition

针对年径流量时间序列非线性、非平稳的问题,采用经验模态分解法(EMD)实现对年径流的多时间尺度、多层次分解,获得简单且平稳性较好的分量,分别选择合适模型对各分量不同变化规律进行合理预测分析,再将各分量预测值结合,重构原始序列年径流预测值,进而建立基于EMD的年径流组合预测模型,并应用于桂江流域年径流预测中。结果表明,该模型预测结果稳定可靠、精度较高,具有推广应用价值。

According to the nonlinearity and non-stationarity of annual runoff,empirical mode decomposition method(EMD) is used to achieve the multi-time scales and multi-level decomposition of annual runoff and simple components with preferable stationarity are obtained.The suitable model for each component is selected to analyze its variation rules.Predicted values of each component are combined to reconstruct predicted values of the original annual runoff sequence.Thus,the model of combination prediction of annual runoff based on the EMD is established.The application results of annual runoff forecasting in Guijiang watershed show that the prediction results are reliable and it has high accuracy,which is worth spreading and application.