exploring the use of optimization models to forecast the penetration rate in blast holes

Using a rotary drilling machine is crucial for exploration and extraction in mining operations. accurately estimating the costs and efficiency of these machines is essential for effective mining and construction project design. the drilling penetration rate (pr) serves as a valuable metric for assessing both the costs and efficiency of rotary drilling rigs. optimizing the pr can lead to reduced costs and enhanced performance of the drilling machine. in this article, we explore an advanced machine learning (ml) approach that integrates light gradient boosting (lgb) and random forest (rf) with meta-heuristic algorithms, such as the grey wolf optimizer (gwo) and harris hawk optimizer (hho), to estimate pr. to validate the proposed models, we applied them to datasets we collected, which include seven independent variables: drilling diameter (inches), rotational speed (rpm), weight on bit (wob), uniaxial compressive strength (ucs) of rock (megapascals), tensile strength (t) of rock (megapascals), joint spacing (js) in centimeters, and relative accuracy of joints (jd) in degrees. the results demonstrate that the rf-gwo method outperforms the others, achieving an r² value of 0.987 and an rmse of 3.059, compared to lgb-gwo (0.912, rmse 8.045), lgb-hho (0.917, rmse 7.831), and rf-hho (0.912, rmse 8.044) on the training data. additionally, sensitivity analysis reveals that the rpm variable significantly impacts pr, while the jd parameter has the least effect.