Landslide Displacement Prediction Model Using Time Series Analysis Method and Modified LSTM Model

Landslides are serious and complex geological natural disasters that threaten the safety of people’s health wealth worldwide. To face this challenge, a landslide displacement prediction model based on time series analysis modified long short-term memory (LSTM) is proposed in paper. Considering data from different periods have values, weighted moving average (WMA) method adopted to decompose cumulative into trend periodic displacement. predict trend, we combined landslides early stage with an LSTM model. repeatability periodicity rainfall reservoir water level every cycle, fully connected (LSTM-FC) was constructed by adding layer traditional The two predicted displacements were added obtain final In paper, under same conditions, used polynomial function algorithm compare LSTM-FC eight other commonly algorithms. Two results indicate able effectively