Abstract: To timely, accurately, and comprehensively acquire the uniaxial saturated compressive strength of rocks at the tunnel face and provide an important prerequisite for promoting intelligent tunnel construction and achieving dynamic changes in rock mass classification, a case study is conducted on the Lingshangcun tunnel of the ChizhouHuangshan highspeed railway. Rockdrilling jumbo parameters from 41 sets of tunnel face and surrounding rock strength data from four sets of rocksimilar materials are obtained, respecrtively. After analyzing the rockbreaking process of the rockdrilling jumbo, four indicators, impact pressure, propulsion pressure, rotation pressure, and drilling rate, are selected as input parameters for the prediction model. Strength prediction models are established based on singlefactor regression, multiple nonlinear regression, and the least squares support vector machine (LSSVM) algorithm. The prediction results of the model are as follows: (1) The singlefactor prediction model, using drilling rate as the input parameter, achieves the highest prediction accuracy with a correlation coefficient of 0.945 8 between the predicted and actual values. (2) The multiplefactor prediction model improves the prediction accuracy further, with a correlation coefficient of 0.974 5. (3) The LSSVM algorithm models demonstrates a correlation coefficient of 0.983 8 between the predicted and actual values, with a prediction accuracy close to that of multiple nonlinear regression. This indicates that both models can be used for rock strength prediction in practical applications.

Key words: rock drilling jumbo, uniaxial saturated compressive strength of rock, drilling parameters, regression analysis, machine learning