Prediction of Landslide Displacement using BP Neural Network Model: A Case Study in Gansu, China

Landslides pose significant threats to the human living environment and severely endanger the life and property safety of surrounding residents. Therefore, conducting landslide displacement prediction and implementing corresponding prevention and control measures are crucial for mitigating such losses. In this study, taking the Beishan Landslide in Gansu Province as a case study, we established a landslide displacement prediction model based on a backpropagation (BP) neural network. We compared the prediction accuracy of two models: a cumulative displacement prediction model that considers only the deformation trend of previous cumulative displacement, and a multi-source data prediction model that integrates rainfall meteorological conditions with previous displacement data. The results demonstrate that the multi-source data prediction model achieves higher accuracy and precision. Specifically, its relative prediction error is less than 0.5%, and it exhibits high prediction effectiveness for linear, concave-down, and concave-up deformation curves.

Keywords: Landslide displacementprediction, BP neural network, Cumulative displacement prediction model, Multi-source data prediction model.