Bi-LSTM-GPR algorithms based on a high-density electrical method for inversing the moisture content of landslide

摘要

Soil moisture content is an essential indicator for landslide monitoring and early warning. A hybrid model of bidirectional long short-term memory cyclic neural network and Gaussian process regression (Bi-LSTM-GPR) based on a high-density resistivity method was proposed to invert the moisture content of landslides. The model was trained through laboratory tests, and the resistivity and water content data monitored in large-scale landslide model tests were used to verify and evaluate the model. The Bi-LSTM-GPR was applied to the inverse moisture content in the field experiment of the Bazimen landslide. The prediction results of the proposed method are in good agreement with the field sampling values, indicating that the method can be applied to the actual monitoring of landslides in the reservoir area.

关键词