Estimating the penetration rate of the machine into the rock is the first and most important step in predicting the time of mechanized tunnel excavation. In the last few decades, many studies have been conducted to predict TBM penetration, in which different methods have been used. In this study, the different methods used to propose a relationship for predicting the penetration rate were implemented to compare their strengths and weaknesses. For this purpose, a database of information during the excavation of the Golab tunnel was created, including machine operating parameters and geomechanical rock parameters. Using the information in the created database, the effect of different geomechanical parameters on machine performance was investigated, and the effect of joint spacing and RQD was greater than other parameters, and uniaxial compressive strength had a small effect on the amount of penetration. Using different regression analysis methods, relationships were presented to predict the penetration rate, and the SMo regression method showed higher accuracy. The PSO algorithm was also used to determine the coefficients of parameters affecting machine performance. In addition, the artificial neural network method was also used, which although had higher accuracy than other methods, was less efficient than other methods due to the lack of a specific relationship. With the help of machine learning methods and decision tree construction, and by prioritizing more effective geomechanical parameters, a classification system for predicting the infiltration rate was proposed. According to the results obtained and the comparison of the methods used, the decision tree method showed the best efficiency, and the result was proposed as a classification system for predicting the infiltration rate.