基于VMD-CNN-LSTM模型的渭河流域月径流预测

Monthly runoff prediction of Weihe River Basin based on VMD-CNN-LSTM model

为了提高月径流预测的精度,解决月径流的非线性和不稳定性对于预测结果的影响,提出了变分模态分解(VMD)、卷积-长短期记忆神经网络(CNN-LSTM)组合的VMD-CNN-LSTM模型。通过选取渭河流域张家山站和魏家堡站1960—2005年的实测月径流数据进行训练和测试,并与EMD-LSTM、EMD-CNN-LSTM、VMD-LSTM模型进行对比,选取均方根误差(RMSE)、平均绝对误差(MAE)和Nash系数(NSE)对测试集的预测结果进行评价。研究结果表明VMD-CNN-LSTM模型相较于其他模型有更优的预测精度以及更低的误差,且对于峰值谷值的拟合更精确,为月径流的预测提供了一条新的途径。

In order to improve the accuracy of monthly runoff prediction and to solve the influence of the non-linearity and instability of monthly runoff on the prediction results,a VMD-CNN-LSTM model combining variational modal decomposition(VMD)and convolution-long short-term memory neural network(CNN-LSTM)is proposed.Training and testing are carried out by selecting the measured monthly runoff data from Zhangjiashan Station and Weijiabao Station in the Weihe River Basin from 1960 to 2005,comparing with EMD-LSTM,EMD-CNN-LSTM,VMD-LSTM models,selecting root mean square error(RMSE),mean absolute error(MAE)and Nash coefficient(NSE)to evaluate the prediction results of the test set.The results show that the VMD-CNN-LSTM model has better prediction accuracy and lower error than other models.The fitting of peaks and valleys is more accurate,providing a new way for the prediction of monthly runoff.