Hybrid ANN–GWO Algorithm for Improving Methane Drainage Efficiency in Cross-Measure Borehole in Underground Coal Mines

Methane release during mining exploitation represents a severe threat to miners’ safety. Methane gas can be removed from coal seams using predrainage and postdrainage techniques and used to generate electricity or for household purposes. One of the postdrainage methods is the cross-measure borehole method, which involves drilling boreholes from the tailgate roadway to an unstressed zone in the roof or floor layers of a mined seam. To achieve high efficiency of gas draining, predicting and determining the optimum range of design elements in gas drainage operation is necessary. The distance between the methane drainage (MD) stations is one of the most important parameters for determining the amount of gas removed. This study was conducted on the basis of the measurement data of MD in the Tabas coal mine. In this study, hybrid multilayer perceptron (MLP) neural networks and gray wolf optimizer (GWO) algorithm were employed to predict and optimize the gas drainage process. Therefore, the technical parameters of MD boreholes, including panel properties, advanced speed, and joint density, were considered. Then, a hybrid artificial neural network (ANN)–GWO algorithm was developed in the MATLAB programing environment, and the performance of the model was evaluated using statistical criteria such as regression relationships, correlation coefficients between actual and predicted values, and average relative error percentage. Applying the presented model can increase the MD efficiency to an acceptable range. The results showed an average reduction in the distance between the MD stations from 20 to 10 m (assuming that other technical parameters of the boreholes remain constant in the cross-measurer boreholes method), and the MD efficiency increases by ~20%–50%. Finally, the efficiency of gas output from the mine will be improved, and a balance will be struck between methane removed with ventilation air methane (VAM) and MD.