Research on RNN and LSTM Method for Dynamic Prediction of Landslide Displacement

Abstract

Good landslide displacement prediction is an important part of the landslide disaster warning.Limited by the nonlinear dynamic characteristics of landslide displacement evolution,historical data is generally missing in traditional prediction methods,resulting in low prediction accuracy.To this end,this paper proposes a deep learning method for landslide displacement prediction to establish two dynamic displacement prediction models of recurrent neural network (RNN) and long short term memory network (LSTM) for comparison,and selects the displacement changes of multiple monitoring points for dynamic prediction by the method of “circulation training”,taking the Xintan landslide project as an example.The results show that when the error function meets the expected accuracy,the LSTM model has higher prediction accuracy.In addition,various evaluation indicators also show that the overall prediction effect of the LSTM model is better.

Keywords

• landslide disaster;displacement prediction;recurrent neural network;long short term memory network;dynamic prediction