Research on the prediction of mining efficiency of thin coal seam working face based on the Random Forest Algorithm

Thin coal seam mining is harsh, whose mining efficiency is affected by a variety of factors in the underground. In order to study the influence of each factor on the mining efficiency and predict the mining efficiency of thin coal seam, fifteen factors, coal seam thickness, coal seam average dip angle, coal seam hardness, coal seam fault, coal seam top and bottom plate, coal seam hydrology, coal seam gangue, gas mine type, coal seam spontaneous combustion, coal seam stability, coal seam undulation, coal mining technology, monthly production, working face length and working face equipment, are extracted from 130 mined thin coal seam working faces. Then, 37 workface samples with complete data are screened, and the optimal prediction model for mining efficiency of thin coal seam working face is trained based on Random Forest Algorithm, Grid Search Method and 3-Fold Cross-Validation Method, with an accuracy rate of about 95.2% and R² of 0.924 in the test set. Meanwhile, the optimal model draws a weighting diagram of mining efficiency, where monthly production, coal mining equipment, working face length and coal seam thickness have a greater influence on mining efficiency, while other parameters have a smaller influence. Finally, taking a thin coal seam working face in Shuangyang Coal Mine as an example, the mining efficiency is predicted, and the predicted values are theoretically verified in terms of working face production, workers’ work efficiency and equipment utilization rate. The results are within the prediction range of the Random Forest Model, which verifies the reliability of the Random Forest mining efficiency prediction model and provides a reference for the mining design of thin coal seam working face.