The invention relates to a method of solving unbalanced distribution of class in excavation of abnormal power consumers. The method of solving unbalanced distribution of class in excavation of abnormal power consumers includes an SMOTE algorithm and an ROSE algorithm, and has the advantages of starting with a training set, changing distribution of the training data by resampling, that is, constructing a synthesis data set for which the sample number is relatively balanced; and modeling on the new data set so as to improve the performance of a classifier and solve the problem of unbalanced distribution of class.