摘要
在地铁线路道岔转辙机工作过程中,需要定期检测其动静接点的接触状态,及时发现和排除工作状态的异常。对道岔转辙机动静接点接触状态的现有检测方法是依靠人工肉眼识别。该方法存在测量偏差大、容易漏警、实时性不强等不足。提出一种基于图像处理和深度学习的道岔转辙机动静接点状态检测技术,在采集大量道岔转辙机动静接点图像的基础上,利用神经网络深度学习方法,对机器进行大量的训练,找出检测点的特征参数;再结合图像处理方法,得到所需的测量参数;同时结合智能手机应用程序,实现对道岔转辙机动静接点的实时检测。该方法可以实现对道岔转辙机动静接点状态的便捷、快速、准确检测,提升对动静接点状态判断的准确性,从而提高道岔转辙机的维护效率和维护质量,确保列车的安全运行。
In the working process of switch machine,regular detection of the movable/static contact state is required,so as to timely find and eliminate problems.The current method to detect the movable/static contact state is based on artificial visual recognition,which has many disadvantages like inaccurate measurement,alarm failure and poor real-time performance.In this paper,a movable/static contact state detection technology based on image processing and deep learning is presented.Based on a large number collection of movable/static contact images,the neural network deep learning method is used for a lot of machine training to find out the characteristic parameters of the detection points.Then,combined with the image processing,the required measurement parameters are obtained.At the same time,by combining with mobile app,the real-time detection of switch machine movable/static contact is realized.This technology can realize convenient,rapid and accurate detection of movable/static contact state,improve the judgement accuracy of the movable/static contact state,the efficiency and quality of switch machine,and ensure the safe operation of trains.
作者
施聪
SHI Cong(Telecom and Signal Branch,Shanghai Metro Maintenance Support Co.,Ltd.,200235,Shanghai,China)
出处
《城市轨道交通研究》
北大核心
2020年第S02期149-152,共4页
Urban Mass Transit
关键词
地铁
道岔转辙机
动静接点
神经网络
图像处理
metro
turnout switch machine
movable/static contacts
neural network
image processing