| VMD-LSTM模型对不同预见期月径流的预测研究 |
| |
| 引用本文： | 祁继霞,粟晓玲,张更喜,张特.VMD-LSTM模型对不同预见期月径流的预测研究[J].干旱地区农业研究,2022,40(6):258-267. |
| |
| 作者姓名： | 祁继霞 粟晓玲 张更喜 张特 |
| |
| 作者单位： | 西北农林科技大学水利与建筑工程学院，陕西 杨凌 712100;西北农林科技大学水利与建筑工程学院，陕西 杨凌 712100；西北农林科技大学旱区农业水土工程教育部重点实验室，陕西 杨凌 712100 |
| |
| 基金项目： | 国家自然科学基金项目（51879222，52079111） |
| |
| 摘 要： | 为减小径流时间序列的非线性及非平稳性带来的预测误差，提高多种预见期下的月径流预测精度，将变模态分解（VMD）和长短期记忆神经网络（LSTM）模型相结合，建立了VMD-LSTM组合预测模型，并将大气环流因子作为模型输入的增加项，预测未来1~3个月的月径流。将模型应用于黄河流域上游唐乃亥、民和、享堂、红旗及折桥站的月径流预测以验证模型的适用性，并与VMD-BP（BP神经网络）、VMD-SVR（支持向量回归）及单一LSTM模型相比较。结果表明：VMD-LSTM组合模型的预测误差最小、精度最高，相比单一LSTM模型，其纳什效率系数（NSE）约从0.6~0.7提高到0.9以上；融合大气环流因子后VMD-LSTM模型预测精度进一步提高，NSE保持在0.91~0.96之间；随着预见期的增长，VMD-LSTM模型预测精度衰减较VMD-BP和VMD-SVR模型明显变缓，在3个月预见期时NSE仍能保持在0.84~0.95之间。VMD-LSTM模型是月径流预测的一种有效方法，结果可为研究区月径流预测提供参考。
|
| 关 键 词： | 月径流预测 变模态分解 长短期记忆神经网络 大气环流 预见期 |
Research on monthly runoff prediction of VMD-LSTM model in different forecast periods |
| |
| QI Jixi,SU Xiaoling,ZHANG Gengxi,ZHANG Te.Research on monthly runoff prediction of VMD-LSTM model in different forecast periods[J].Agricultural Research in the Arid Areas,2022,40(6):258-267. |
| |
| Authors: | QI Jixi SU Xiaoling ZHANG Gengxi ZHANG Te |
| |
| Abstract: | In this study, Variational Mode Decomposition (VMD) and Long Short\|Term Memory (LSTM) neural network were integrated to establish a hybrid prediction model. The named VMD-LSTM model was to reduce the prediction error caused by the nonlinearity and non\|stationarity of the runoff series and improve the accuracy of monthly runoff prediction results under various forecast periods. Some highly correlated atmospheric circulation factors were selected as the additional term of the model input to predict the monthly runoff for 1~ 3 lead months. The performance of VMD-LSTM in predicting monthly runoff at the Tangnaihai, Minhe, Xiangtang, Hongqi and Zheqiao stations at the upper reaches of the Yellow River Basin was verified. The VMD-LSTM model was compared with VMD-BP (BP neural network), VMD-SVR (support vector regression) and the single LSTM model for evaluating its applicability.The results showed that the VMD-LSTM model exhibited the best forecasting performance, compared with the single LSTM model, and its Nash efficiency coefficient (NSE) was substantially improved from 0.6~0.7 to above 0.9. When putting atmospheric circulation factors, the accuracy of VMD-LSTM model was further improved, with NSE remaining at 0.91~0.96. With the increase of lead time, the precision attenuation of VMD-LSTM model became slower than VMD-BP and VMD-SVR model, and its NSE still remained at 0.84~0.95 when the forecast period was 3 months. The VMD-LSTM model is an effective method for monthly runoff prediction, and the results can provide guidance for monthly runoff prediction in the study area. |
| |
| Keywords: | monthly runoff prediction variable modal decomposition long short\|term memory neural network atmospheric circulation forecast period |
|
| 点击此处可从《干旱地区农业研究》浏览原始摘要信息 |
|
点击此处可从《干旱地区农业研究》下载全文 |
