Abstract: The phenomenon of CO₂ corroding underground pipe is common, but there is no suitable method to predict the corrosion rate of 13Cr pipes in gas-phase CO₂ environment. To solve the problem, the corrosion resistance mechanism and influencing factors of pipes containing Cr were investigated. The importance of each corrosion factor affecting the corrosion rate of 13Cr pipes based on the orthogonal experiment method was determined. A prediction model of pipe corrosion rate was established based on machine learning method. Four kinds of machine learning models, including BP Neural Network,Decision Tree, Random Forest and Support Vector Machine, were compiled by Python language, and the corrosion rate model was optimized by using laboratory experiment data as data set. The results show that the corrosion of 13Cr pipe is uniform in the gas phase CO₂, and the corrosion rate is 0.0122 mm/a-0.0361 mm/a. Compared with the experimental results, the prediction accuracy of the decision tree model can reach 85.1%. The remaining casing wall thickness was used to verify the model, and the accuracy was over 80%. The established prediction method can be applied to the corrosion rate prediction of pipes under similar corrosion conditions.