Pantograph structure abnormal dynamic video monitoring method and device

Abstract

The invention relates to a pantograph structure abnormity dynamic video monitoring method and device. The method comprises the following steps: S1, obtaining a training set monitoring video, extracting pantograph pictures, recognizing a pantograph and a pantograph support, and marking each pantograph picture according to the recognized pantograph and pantograph support; S2, taking the marked pantograph picture as training set data, and establishing and training an SSD target detection model; and S3, collecting a pantograph monitoring video, extracting a pantograph picture based on the collected pantograph monitoring video, and inputting the obtained pantograph picture into the trained SSD target detection model to obtain a detection result. Compared with the prior art, the generalization and stability of detection are improved.

Description

technical field [0001] The invention relates to the field of fault detection of pantographs of electric locomotives, in particular to a method and device for dynamic video monitoring of pantograph structural abnormalities. Background technique [0002] The pantograph is located on the roof of the electric locomotive and extracts energy from the catenary to provide electrical energy for the vehicle. During the operation of the locomotive, the pantograph may have structural abnormalities, foreign matter intrusion, pantograph-catenary sparks, and eccentric wear of the skateboard. Since current electric locomotives are powered by high current, problems with the pantograph will affect the life and performance of the equipment, cause vehicle damage and system failure, and cause greater losses. [0003] The existing methods for image detection of pantograph-related faults through video mainly use the YOLO algorithm and the Faster R-CNN algorithm. Although the YOLO algorithm has a...

AI Technical Summary

Problems solved by technology

Although the YOLO algorithm is fast, it has the disadvantage of low positioning accuracy, and it is difficult to detect small targets; although the Faster R-CNN algorithm can improve the accuracy, it has the disadvantage of slow detection speed

[0004] The use of algorithms and training models in the existing pantograph fault video surveillance system is generally established and trained directly in the development board or video surveillance system, which has the following defects: the processing speed and storage capacity of the development board or video surveillance system Space and other performance are weak compared with computers, which limits the training speed and detection accuracy of the model

Images