Abstract: In order to improve the prediction accuracy of soil corrosion rate of buried pipelines, the influencing factors of soil corrosion were sorted out and analyzed. The data dimensionality of influencing factors was reduced by Kernel Principal Component Analysis (KPCA), the key parameters of the Support Vector Machine (SVM) were optimized, and comparison was made among the four models of GWO-SVM, GA-SVM, PSO-SVM and FOA-SVM. The results show that the KPCA model can effectively reduce the dimensionality of the prediction model, in which the five factors, i.e., the soil resistivity, REDOX potential, salt content, Cl^- mass fraction and water content, have a great influence on pipeline corrosion. In addition, GWO-SVM has the smallest mean absolute error and root mean square error, which are 1.90% and 0.098 909, respectively, and it takes the minimum training time (2.55 s) among the four models. Therefore, the KPCA-GWO-SVM model is better for the prediction of soil corrosion rate of buried pipelines, and the research results could provide theoretical basis and practical reference for pipeline integrity management.