6th INTERNATIONAL SYMPOSIUM ON MINE SAFETY SCIENCE AND ENGINEERING

158 / 2023-07-05 19:39:14

Prediction of Gas Emission in Working Face Based on LASSO-WOA-XGBoost

Gas emission prediction; Least Absolute Shrinkage and Selection Operator (LASSO); Whale Optimization Algorithm (WOA); eXtreme Gradient Boosting (XGBoost) ; Safety engineering

Abstract:With the increasing intensity and depth of coal mining, gas outbursts have become more severe. To improve the accuracy of gas emission prediction in the mining working face and selecting the LASSO (Least Absolute Shrinkage and Selection Operator), Whale Optimization Algorithm (WOA), and eXtreme Gradient Boosting (XGBoost) algorithms were used to construct the LASSO-WOA-XGBoost gas outburst prediction model. Thirteen factors affecting gas emission were selected using LASSO for feature selection out of which nine factors having higher effects on gas outburst were considered. WOA optimized the three main parameters in XGBoost, namely, n_estimators, learning_rate, and max_depth. This helped to improve the prediction effect of XGBoost algorithm and solved the problem of difficulties in parameter tuning due to many parameters of the algorithm. Comparative analysis of LASSO-XGBoost and PCA-WOA-XGBoost prediction models showed that LASSO feature selection is better than PCA dimensionality reduction in improving the accuracy of the prediction model. In comparison to the other two prediction models,the LASSO-WOA-XGBoost model had an mean absolute error of 0.1775 and a root mean square error of 0.2697. indicating higher prediction accuracy. This method provides a new approach for predicting the gas outburst in the mining production of the working face.