Prediction of gas hazard in coal stratum tunnels based on improved snake optimizer and support vector machine

Abstract

In tunnel engineering that passes through coal-bearing strata, gas explosion accidents pose a severe threat to the safety of construction personnel. Therefore, accurately predicting gas risks during the planning and design stages of tunnels is crucial. This paper proposed a gas hazard prediction method based on support vector machine (SVM) with improved snake optimizer (ISO) for more accurate prediction and classification of hazard levels. Firstly, five improvement strategies were adopted to enhance the global search capability and robustness of snake optimizer (SO). The nine testing functions were used to comprehensively test, compare, and analyze the performance of the ISO with other optimization algorithms. Then, the model was used to learn and test from a database of 80 collected coal gas tunnel cases, on which the ISO-SVM gas outburst prediction model was established. The improved snake optimizer algorithm significantly boosted the classification performance of the Support vector machine, achieving a test set prediction accuracy of 93.8%. The validated model was applied to four newly constructed tunnel projects in Sichuan and Yunnan Provinces, China, and the prediction results were consistent with the actual hazard levels. Compared to traditional methods, the proposed model overcomes the limitations of single-indicator determination and effectively addresses the issue of poor applicability in gas outburst determination due to potential data deficiencies. In addition, a comprehensive comparison was conducted with other machine learning models, and the ISO-SVM prediction model demonstrated superior predictive performance, highlighting its outstanding potential and practical applicability in future gas hazard prediction.