Landslide risk prediction by using GBRT algorithm: Application of artificial intelligence in disaster prevention of energy mining

Geological disasters on the slopes of open-pit mine dumps in energy extraction fall into the category of mine production process safety. For the mine safety, it is crucial to accurately predict the landslide risk of open-pit mine dumps. In order to prevent landslide geological disasters in open-pit mine dumps under the effect of heavy rainfall, this study establishes a fast and accurate landslide risk prediction model for open-pit mine dumps based on machine learning (ML). Given the actual geological conditions and rainfall of the slope of an open-pit mine dump in Shaanxi Province, Geo-Studio software is used to calculate the factor of slope safetyunder different states, and the gradient boosting regression tree (GBRT) algorithm model is used to predict the factor of slope safety. The comparison with the prediction results of different algorithms shows that the GBRT model has the highest prediction accuracy; meanwhile, the GBRT model predicts the factor of safety (FOS=1.283) for the bench slope of the dumps under the rainfall intensity (q=87 mm/d) of the 20-year rainstorm recurrence period , and its error is smaller than that calculated by the numerical simulation analysis (FOS=1.289). Therefore, the GBRT model has better applicability in predicting the safety factors of open-pit mine dumps slope under the effect of heavy rainfall, which is of great significance to realize the early warning of landslide risk in open-pit mine dumps.

Automated summary

This study establishes a fast and accurate landslide risk prediction model for open-pit mine dumps under the effect of heavy rainfall based on machine learning (ML). The GBRT model predicts the safety factors of open-pit mine dumps slope under the effect of heavy rainfall, which is of great significance for early warning of landslide risk in open-pit mine dumps.