Research on prediction model of coal mine roadway surrounding rock stability based on ISCSO-KELM

To accurately predict the stability of coal mine roadway surrounding rock, a prediction model based on an Improved Sand Cat Swarm Optimization (ISCSO) algorithm optimizing Kernel Extreme Learning Machine (KELM) is proposed. The model incorporates Bernoulli chaotic mapping, Lévy flight, Symbiotic Organisms Search (SOS), and Random Opposition-Based Learning (ROBL) strategies to enhance the global exploration and local exploitation capabilities of the sand cat swarm algorithm. Subsequently, ISCSO is utilized to optimize the penalty coefficient and kernel parameters of KELM, constructing the ISCSO-KELM prediction model. An evaluation index system is established based on surrounding rock strength, in-situ stress, rock mass integrity, and mining-induced effects, and their correlations are analyzed. Experimental results demonstrate that ISCSO-KELM outperforms comparative models in terms of Accuracy, Macro-Precision, Macro-Recall, and Macro-F1. Engineering validation in the Sanjiaohe Mine and Mugua Mine shows that the prediction results align with the actual conditions, confirming the model’s effectiveness and applicability.