Critical early warning of underground debris flows in mines based on rainfall–collapse characteristics

The natural caving method of mining is susceptible to underground debris flow disasters under the combined influence of high-intensity rainfall and surface subsidence. Research on early warning schemes for underground debris flows utilizes characteristic parameters such as the effective cumulative rainfall in the early stage, the daily triggered rainfall, and the collapse area as the analysis indices. The random forest algorithm is modified based on variable weight theory to obtain the index weight, and an occurrence probability and hazard analysis of underground debris flows is conducted via the Technique for Order Preference by Similarity to Ideal Solution and the Vlsekriterjumska Optimizacija I Kompromisno Resenje combination evaluation. The comprehensive hazard evaluation factor is used to obtain the characteristic index threshold under different hazard levels, thus achieving hierarchical early warning of underground debris flows in mines. The results reveal that hazard comprehensive evaluation factors of 0–0.2, 0.2–0.5, and > 0.5 correspond to low, moderate, and high hazards, respectively. The hazard level under the coupling condition of the mine global characteristic parameters is given, and an early warning method is established to prevent underground mine debris flows. This evaluation method has implications for early warning of underground debris flows in mines, thus preventing disasters and ensuring production safety.