Abstract: The rate of penetration of down-the-hole（DTH） drilling rig is recognized as a pivotal metric for gauging operational efficiency, which is integral to scientifically devising work strategies and accurately estimating excavation costs. This paper focuses on predicting the rate of penetration of open-pit DTH drilling rig. By conducting in-situ drilling measurement experiments at open-pit mining sites, it establishes a predictive model for rate of penetration using an enhanced XGBoost algorithm. The results indicate that the model achieved an RMSE of 0.001 27 and an R² value of 0.909 on the test set, outperforming light gradient boosting machines, support vector machines and neural networks. Specifically, the RMSE is reduced by 24.9%, 24.0%, and 30.2%, respectively, while the overall R² improved by 12.8%, 16.4%, and 22.3% respectively. When predicting the rate of penetration for diorite, the enhanced XGBoost model demonstrates a pronounced improvement in predictive performance, and its accuracy decrease is relatively moderate as the actual rate of penetration increases. Within this model, the torque and impact air quantity are identified as the two most critical factors influencing the DTH rate of penetration, whereas under the experimental conditions of this study, the rotation speed and impact power exhibites a relatively diminished influence. This study further proposes a working parameter combination based on the established model for the drilling rig, and it proves to reduce the fuel consumption by 14.6%.