Research On Key Technologies Of Intelligent Image Processing Of High-speed Railway Pantograph Arc And Structural Anomaly

With the rapid development of China Railway Highspeed,higher monitoring requirements have been put forward for the traction power supply system.Pantograph is a key component in the power supply system of high-speed trains,the damage and failure of pantograph will seriously affect the safety of high-speed trains operations.During the long-term operation of high-speed trains,the focus of detection objects are the arc caused by poor pantograph-catenary contact and the structural abnormality caused by pantograph deformation,while the traditional detection method can’t meet the requirements of online pantograph monitoring.Aiming at the detection problem of pantograph arc and structural abnormality,a detection method based on deep learning YOLOv3 object detection algorithm and image processing technology is proposed.Firstly,identify and locate objects in the onboard pantograph monitoring camera by using the deep learning YOLOv3 object detection algorithm.Due to the different detection objects,a pantograph arc detection model and a pantograph structural anomaly detection model are designed respectively.In the pantograph arc detection model,an image segmentation method based on visual saliency is proposed to extract the arc or strip contour.In the pantograph structure anomaly detection model,the improved Otsu algorithm is used to obtain the pantograph contour,and the upper edge extraction method of pantograph based on the Harris algorithm is used to obtain the pantograph upper edge curve,and the pantograph state is judged according to the similarity of the pantograph upper edge curve to achieve the detection of structural abnormal state.The test experiment of pantograph arc and structural abnormality completed by simulating the pantograph detection during the real operation of high-speed trains.The experiments show that the algorithm model designed can meet the requirements of pantograph arc and structural anomaly detection for high-speed trains in terms of accuracy and real-time performance.