Research on height prediction of water-conducting fracture zone in coal mining based on intelligent algorithm combined with extreme boosting machine

Abstract

The development of water-conducting fracture zones (WCFZ) caused by coal mining is a destructive phenomenon of mining damage. The development height of the WCFZ is an important reference index for the prevention and control of safety hazards such as water infiltration or sand intrusion in mines and for program optimization. In this work, a combinatorial optimization model for height prediction of WCFZ is established by combining the extreme gradient boosting machine and several commonly used intelligent algorithms: genetic algorithm, particle swarm optimization, jaya algorithm, sparrow search algorithm. Accordingly, 142 sets of WCFZ composed of 4 main parameters were selected as the input independent variables, and the maximum height (H) was selected as the output dependent variable. In addition, to demonstrate the validity of the proposed combinatorial optimization models, this study used the following four models (Random Forest, Bagging, XGBoost and AdaBoost) for comparison. Model performance evaluation criteria including the coefficient of determination, root mean square error, mean absolute error, and variance accounted for are used to evaluate the model. In this work, 142 sets of cases to the WCFZ analyzed and the method of Shapley Additive Explanations was used to explain the importance and contribution of features to maximum height prediction. The research results show that, compared with other machine learning models, the combinatorial optimization model put forward in this work can improve the predicted accuracy and reliability of height prediction of WCFZ. The research in this work may help researchers use machine learning models to predict and research the height development of WCFZ.