Dam deformation prediction model based on CNN-Attention-LSTM
[Objective]Predicting dam deformation to avoid risks is the focus of dam deformation monitoring,and a reliable predicting model can provide insights into the future deformation trend of the dam.In order to better predict the deformation of the dam and improve prediction accuracy and calculation efficiency,[Methods]this paper proposes a dam monitoring model based on Convolutional Neural Network(CNN),attention mechanism(Attention)and Long Short-Term Memory network(LSTM).CNN extracts features from monitoring data,and LSTM learns better from time series data.Based on this CNN-LSTM model,Attention mechanism,one type of deep learning algorithm,is coupled to highlight the impact of features on the input effect without affecting the accuracy of the model.On the premise of improving the calculation speed,further improving the model prediction accuracy and stability.Through engineering example analysis,[Results]the model proposed in this arti-cle can accurately predict dam deformation.The average R2,MAE,RMSE,MSE and MAPE on the test set at each point are 0.989,0.337 mm,0.469 mm,0.252 mm and 13.918％respectively.[Conclusion]The result show that the built model has better deformation prediction ability.Compared with CNN,LSTM,CNN-LSTM,and Attention-LSTM models,this model has better MAE,RMSE,MSE,MAPE,and R2,etc.indicators,and improves computational efficiency,making it more suitable for the prediction of dam deformation.
万方数据
维普
