Abstract: As a key part of the power system,substation equipment is of great significance to maintain its safe and stable operation. When the substation equipment fails,the fault type needs to be diagnosed timely and accurately. Aiming at this problem,this paper proposes a fault diagnosis method for substation equipment based on image processing and semisupervised learning. First,we execute feature extraction process on the collected infrared image data, and extract the temperature features,texture features and shape features as the model reference vectors. Then, the synthetic minority oversampling technique（SMOTE）algorithm is used to expand the sample of the minority samples with the label. Finally,we aggregate the unlabeled sample data to construct a graph semisupervised learning network and then train this graph. Compared with traditional supervised learning methods, the proposed method in this paper can learn information from unlabeled sample data. Finally,we test our proposed method on real dataset. The experimental results show that the use of feature extraction,sample generation and semi-supervised learning model can improve the accuracy of substation equipment fault classification,which verifies the effectiveness of the method proposed in this paper.