关键词: 经验模态分解, 长短期记忆神经网络, 渗流, 大坝安全监测

Abstract: To improve the accuracy and efficiency of seepage prediction for earth rock dams, a seepage prediction model based on Empirical Mode Decomposition and Long Short Term Memory is proposed in this paper. Regarding the obtained seepage monitoring time series data of earth rock dam, the EMD method is used to adaptively decompose them and obtain a number of intrinsic mode functions from high frequency to low frequency. The LSTM earth rock dam seepage prediction model is established for intrin ⁃ sic mode functions, and the sum of prediction results of each intrinsic mode function is the final predic ⁃ tion value of the earth rock dam seepage. To verify the accuracy of this earth rock dam seepage predic ⁃ tion model, the seepage data of an earth rock dam is analyzed and compared with other prediction mod ⁃ els through experiments. The results show that the seepage prediction model of earth rock dams based on EMD- LSTM has good accuracy and is applicable to the prediction and analysis of earth rock dam seepage during operation. The results can provide technical support for the decision-making of safe op ⁃ eration of earth rock dams.

Key words: Empirical Mode Decomposition, Long Short Term Memory, seepage, dam safety monitoring