Landslide Displacement Prediction Based on Variational Mode Decomposition and GA–Elman Model

Abstract

Landslide displacement prediction is an important part of monitoring and early warning systems. Effective displacement prediction is instrumental in reducing the risk of landslide disasters. This paper proposes a displacement prediction model based on variational mode decomposition and a genetic algorithm optimization of the Elman neural network (VMD–GA–Elman). First, using VMD, the landslide displacement sequence is decomposed into the three subsequences of the trend term, the periodic term, and the random term. Then, appropriate influencing factors are selected for each of the three subsequences to construct input datasets; the rationality of the selection of the influencing factors is evaluated using the gray correlation analysis method. The GA–Elman model is used to forecast the trend item, periodic item and random item. Finally, the total displacement is obtained by superimposing the three subsequences to verify the performance of the model. A case study of the Shuizhuyuan landslide (China) is presented for the validation of the developed model. The results show that the model in this paper is in good agreement with the actual situation and has good prediction accuracy; it can, therefore, provide a basis for early warning systems for landslide displacement and deformation.