首页 | 本学科首页   官方微博 | 高级检索  
     检索      

编组站提钩自动化智能识别设计
引用本文:安迪,马玉坤,宋海锋,李杨.编组站提钩自动化智能识别设计[J].铁道运输与经济,2020(3):54-60.
作者姓名:安迪  马玉坤  宋海锋  李杨
作者单位:中国铁道科学研究院研究生部;中国铁道科学研究院集团有限公司运输及经济研究所;北京交通大学电子信息工程学院;沈阳奇辉机器人应用技术有限公司
基金项目:国家自然科学基金项目(61903021);中国铁路总公司科技研究开发计划课题(2017X009-A);中国国家铁路集团有限公司科技研究开发计划课题(J2019X004)。
摘    要:为了提升编组站解编效率与现场作业安全性,进一步提高编组站作业自动化技术水平,针对解编自动化技术瓶颈,提出一种基于深度学习的编组站提钩自动化智能识别设计方案。阐述编组站提钩自动化智能识别的设计需求,从位置信息获取、作业过程监控、图像数据采集、信息数据处理等环节研究图像信息处理方法,重点研究智能识别流程,利用深度卷积神经网络模型与算法研究图像特征值提取与图像匹配识别技术实现方案。经验证,智能识别方案在试验中的平均可靠度达99.37%,为编组站解编自动化提供了一种有效技术手段。

关 键 词:编组站  铁路驼峰  智能识别  自动提钩  图像识别  深度学习  可靠度

An Intelligent Recognition Design of Automatic Wagon Uncoupling in Marshalling Yard
Institution:(Postgraduate Department,China Academy of Railway Sciences,Beijing 100081,China;Transportation&Economics Research Institute,China Academy of Railway Sciences Corporation Limited,Beijing 100081,China;School of Electronic and Information Engineering,Beijing Jiaotong University,Beijing 100044,China;QuickHigh Robot Corporation Limited,Shenyang 110027,Liaoning,China)
Abstract:To improve the efficiency and safety of uncoupling work and further improve the automation technology level of marshalling yard, on the basis of expounding the concept of intelligent recognition and deep learning and analyzing the difficulties of automatic wagon uncoupling intelligent recognition, an intelligent recognition system of railway marshalling yard automatic wagon uncoupling is proposed. This paper studies the hardware structure of the system, describes the workflow of image acquisition, image preprocessing, image feature value extraction, and image matching recognition, and studies a deep convolution neural network model and algorithm combined with deep learning technology. After testing, the average reliability of the intelligent recognition system of railway marshalling yard automatic wagon uncoupling in the test is 99.37%, which provides an effective technical means for marshalling uncoupling automation.
Keywords:Marshalling Yard  Railway Hump  Intelligent Recognition  Automatic Wagon Uncoupling  Image Recognition  Deep Learning  Reliability
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号