High-Speed Railway Pantograph-Catenary Anomaly Detection Method Based on Depth Vision Neural Network

A pantograph-catenary system (PCS) is an important part of the high-speed railway power supply system. The quality of the PCS determines the stability of the traction power supply quality of the train. Due to the interaction of the pantograph and the catenary, the uneven distribution of the contact points (CPT) of the pantograph and the contact wire can lead to the failure of the PCS. Therefore, it is necessary to monitor the status of the PCS by detecting contact points. However, the detection frame rate of existing methods is low and the detection accuracy still needs to be improved in complex backgrounds. To solve this problem, this paper proposes a deep visual neural network detection method. The proposed method is divided into two stages. In the first stage, a deep pantograph detection network (DPDN) is established to identify pantograph areas in different complex scenarios. In the second stage, the image visual feature extraction (IVFE) algorithm is used to detect the contact point between the pantograph and the catenary in real time in the pantograph area. Finally, the experimental results demonstrated the speed and the accuracy of the proposed method.