Analysis of rain-induced underground debris flow disasters in mines based on logistic regression models

Abstract Rainfall is one of the main causes of underground debris flow disasters in mining areas. Studying the rainfall threshold required to induce sudden underground debris flow disasters in mines can provide a scientific basis for the prevention and control of such disasters in the rainy season. This paper analyzes the typical cases of underground debris flow disasters caused by rainfall in Pulang mining area from 2020 to 2022. A total of 249 rainfall events were selected, including 86 potential underground debris flow events and 5 major underground debris flow events. The logistic regression model is used to analyze the triggering rainfall before the event and the antecedent effective rainfall. The model evaluation results showed that the accuracy, AUC value and F1 value of the model were 0.85, 0.9493 and 0.85, respectively. The results show that the rainfall threshold of underground debris flow can be divided into three risk levels: P = 0.9: Triggering rainfall 88.6483 mm, antecedent effective rainfall 164.9885 mm. P = 0.7: Triggering rainfall was 78.2563 mm, and the antecedent effective rainfall was 145.6473 mm. P = 0.5: Triggering rainfall was 71.7336 mm, and the antecedent effective rainfall was 133.5076 mm.