Aiming at vegetation shading faults in photovoltaic (PV) power stations, a fault diagnosis model based on support vector machine (SVM) is proposed. According to the seasonal variation characteristics of PV string currents, the characteristics of vegetation shadings are extracted, and the characteristic matrix is established. Through the field operation data, the model was established for training and verification. The diagnostic results of SVM with different kernel functions are compared through experiments, and the diagnostic results of different machine learning are compared. The results show that the SVM algorithm can effectively identify shading faults of PV power stations, and the fault diagnosis accuracy is high.