The rapid measurement and acquisition of the fracture characteristics and strength of the surrounding rock are crucial for evaluating the stability of underground engineering. In this study, a laboratory/coal mine universal drilling test platform was developed to enable real-time measurement of drilling parameters. The testing results demonstrate that under identical rotation speed conditions, both thrust force and torque increase exponentially with increasing drilling speed, while under identical drilling speed conditions, both thrust force and torque decrease linearly with increasing rotation speed. In intact rock layers, variations in torque, thrust force, and drilling speed with depth remain relatively stable, in rock layer interfaces and fragmentation zones, sudden changes occur within a drill displacement range of approximately 3–7 mm, which become more pronounced as strength differences between adjacent rock layers increase.Based on the cutting mechanics model of drill bit, the evaluation index of drilling energy consumption of unit rock mass (Eη) is derived. Subsequently, based on drilling speed, rotation speed, thrust force, torque and energy index, a machine learning model for predicting rock strength is established using the LS-SVM nonlinear regression approach. Field practice has shown that drilling platform can effectively identify both the rock interface and range of fragmentation areas. The experimental results are of great value to the selection of supporting types in the process of coal mine roadway excavation.