Abstract: By the combination of the macro characteristics and acoustic emission characteristics of storage tanks, the macro data processing of storage tanks is guided by the possible corrosion conditions. At the same time, in view of the small number of samples and unbalanced distribution of acoustic emission grading data of storage tanks, a model combining the synthetic minority oversampling technique (SMOTE) and gray wolf optimizer-based support vector machine (GWO-SVM) is used to intelligently predict the safety state grade of storage tanks. The results show that this model can effectively improve the accuracy and reliability of identifying small samples and unbalanced data.