Thermal Hazard Evaluation and Prediction in Deep Excavations for Sustainable Underground Mining

With the advent of the deep mining era, thermal damage in mines has become increasingly significant. The high-temperature environment in underground mines adversely impacts the physiological and psychological health of operators, reduces work efficiency, elevates the risk of accidents, and disrupts sustainable mining operations. Consequently, the development of accurate and reliable methods for classifying thermal hazards is essential for enabling mining enterprises to implement effective prevention strategies. Furthermore, such methods provide a theoretical basis for the sustainable management and utilization of geothermal energy. This study systematically considered factors influencing underground thermal damage and selected 10 quantitative indicators, encompassing both natural and human factors, as evaluation criteria. The CRITIC method was employed to determine the weight of each indicator, which was then integrated with uncertainty measurement theory to develop a novel thermal hazard assessment framework (CRITICUM). This framework enables the classification of thermal hazards in deep mine roadways. The evaluation results generated by the CRITICUM system were subsequently used to train machine learning predictive models. During the training process, the particle swarm optimization algorithm (PSO) was utilized to identify the most suitable prediction model parameters for the complex thermal environment of deep mines by leveraging its capability for continuous iterative evolution. The optimized parameters replaced the original random forest (RF) model parameters, resulting in an enhanced thermal damage prediction model (PSO-RF) with an accuracy of 96.55%, outperforming the standard RF model by 3%. Finally, the prediction model’s accuracy was validated using engineering case data, demonstrating that the results met practical engineering requirements. In summary, the proposed CRITICUM-PSO-RF evaluation and prediction model can accurately classify thermal damage in deep mines and provide a valuable reference for ensuring site safety and supporting the sustainable utilization of geothermal energy.