Novel Intelligent Approach for the Early Warning of Rainfall-Type Landslides Based on the BRB Model

Abstract

This work attempts to apply belief rule-based (BRB) model in information fusion method to landslides, to improve the accuracy and efficiency for early warning of landslides. Taking a typical rainfall-type landslide as the experimental area, the monitoring results find that the surface displacement is the most sensitive monitoring data. It is determined that the monitoring data of surface displacement change rate and rainfall intensity could be used as the input parameter of the BRB model. An initial BRB model is established by setting up the rule base for discriminating warning levels. The data from three monitoring points are collected for the optimization of the initial BRB model, and verification of the optimized BRB models. Results shows the optimized BRB model can accurately describe the nonlinear relationship between the selected monitoring data and the warning level, which provides an intelligent method for landslide prevention and has a strong application prospect.