基于EEMD-GWO-LSTM的滑坡位移预测

Prediction of landslide displacement based on the EEMD-GWO-LSTM

为实现滑坡位移时间序列精准预测,提出一种基于集合经验模态分解(EEMD)数据处理技术和灰狼优化(GWO)算法-长短时记忆网络(LSTM)的精细滑坡位移时序预测模型。该模型采用集合经验模态分解方法将采集得到的累计滑坡位移分解为若干个固有模态函数和1个残余分量,将固有模态函数重构为随机性位移和周期性位移,将残余分量定义为趋势性位移。采用长短时记忆网络对随机性位移和周期性位移进行预测,引入灰狼优化算法对长短时记忆网络模型的超参数进行寻优,建立最优预测模型;依据趋势性位移发展特征,采用最小二乘法实现预测;最后,将这三类位移值进行求和即可得到累计位移预测值。以三峡库区白水河滑坡为例,验证了EEMD-GWO-LSTM模型的可靠性。结果表明:本文提出模型的均方根误差、平均绝对误差和判定系数分别为9.82mm、8.19mm和0.96,均优于BP神经网络和LSTM模型。

To achieve accurate prediction of landslide displacement,a fine landslide displacement time series prediction model based on the ensemble empirical modal decomposition(EEMD)data processing technique and the gray wolf optimization(GWO)algorithm long and short term memory(LSTM)network is proposed.The model uses the ensemble empirical modal decomposition method to decompose the collected cumulative landslide displacements into a number of intrinsic modal functions and one residual component,reconstructing the intrinsic modal functions into stochastic and periodic displacements,and defining the residual component as the trendline displacement.The long and short term memory network is used to predict the stochastic and periodic displacements,and the gray wolf optimization algorithm is introduced to optimize the hyper-parameters of the long and short term memory network model to establish the optimal prediction model.Then,based on the developmental characteristics of the trend displacement,the least squares method is used to realize the prediction.Finally,the accumulated displacement prediction can be obtained by summing the stochastic displacement,periodic displacement and trend line displacement.The reliability of the EEMD-GWO-LSTM model is verified by taking the Baishuihe landslide in the Three Gorges reservoir area as an example.The results show that the RSME,MAE and R2 of the model proposed in this paper are 9.82 mm,8.19 mm and 0.96,respectively,which are better than the BP neural network and the LSTM model.