Abstract: Objective Rail transit vehicle catenary-pantograph system is an important power supply media for electric bus. The structural integrity of its critical component pantograph horn directly impacts train operational safety.Traditional manual inspections face efficiency and real-time limitations, driving the demand for intelligent anomaly detection. Method An intelligent detection method based on deep learning and SVM（support vector machine） techniques is proposed to identify pantograph horn abnormal conditions. After obtaining real-time images of the pantograph horn via onboard catenary operation condition monitoring devices, the YOLOv5 model and U-net model are used for accurate pantograph localization and pantograph horn image segmentation under complex environmental conditions. Subsequently, the extracted features are classified using an SVM algorithm to determine whether the horn is in an abnormal state. Result ＆ Conclusion Validation results show that the proposed method can effectively detect abnormal conditions of the pantograph horn during actual train operation, with both precision and recall rate of detection exceeding 90%. This significantly enhances the efficiency of detecting structural abnormal conditions in pantograph systems and demonstrates promising practical value and potential for widespread application.