Landslide displacement prediction based on Variational Mode Decomposition and Temporal Convolutional Network

Abstract The deformation process of landslides has complex nonlinear characteristics, making the accurate prediction of landslide displacement challenging. To reduce the complex displacement process and better realize the accurate and effective prediction of landslide displacement, this study proposes a hybrid model based on variational mode decomposition (VMD) and a temporal convolutional network (TCN). The VMD method was used to decompose the time series of landslide cumulative displacement data and environmental impact factors (reservoir level and rainfall) and the TCN model to make predictions. To verify the performance of the model, we predicted the cumulative landslide displacements at points ZG93 and ZG118 of the Baishuihe landslide in Zigui County, Three Gorges reservoir area. The results showed that the VMD-TCN model had higher accuracy than other models and can thus play a role in the prediction of landslide displacement and provide a reference for the early warning of landslide displacement deformation.