动车组受电弓状态图像检测技术研究

Research and Implementation of Image Detection System for EMU Pantograph Status

受电弓安装于动车组车顶,其关键零部件安装状态、碳滑板运行状态直接关系到动车的动态运行安全。现有动车车顶的图像检测主要采用线阵相机拍摄,其缺点主要为图像立体感不强、线阵图像存在畸变等。采用一种新的基于深度学习的触发机制,通过高清面阵相机全方位拍摄受电弓,能够更加精准的呈现车顶受电弓全景图像,更加有利于图像智能识别。

A pantograph is installed on the roof of an EMU, and the installation status of the key components of the pantograph and the running status of its carbon contact strips are directly related to the dynamic operation safety of the EMU. At present, image detection of EMU roofs is undertaken by line scan cameras in most cases, and the major shortcomings of this method are that the images do not create strong 3D effects and there is distortion in the line scan images. Therefore, this paper adopts a novel trigger mechanism based on deep learning: high-definition area scan cameras are used to capture the images of pantographs in all directions, which can present the panoramic images of the pantographs more accurately, and facilitate intelligent image recognition.