基于YOLOV5的火车连接手柄识别方法及在自动摘钩机器的应用
DOI:
作者:
作者单位:

1.中电投锦州港口有限责任公司;2.华北电力大学

作者简介:

通讯作者:

中图分类号:

TP391. 41

基金项目:

国家自然科学基金(52275109);河北省自然科学基金(E2022502007)


Train connection handle recognition method based on YOLOV5 and its application in automatic unhook machine
Author:
Affiliation:

China Power Investment Jinzhou Port Co., Ltd

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    能否准确识别火车车钩的连接手柄对摘钩机器人的工作具有重大意义。针对这一问题,提出一种基于改进YOLOv5的火车连接手柄目标识别算法,在主干特征提取网络中使用SAC模块替换C3模块,并引入WIOU(wise intersection over union)函数作为新的边界框损失函数,增强主干网络的特征提取能力,提高模型的泛化能力和收敛速率。然后采集生产现场不同环境、不同位置下的火车车钩连接手柄图片进行识别,实验结果表明改进的YOLOv5算法对火车车钩连接手柄的目标识别率达到97.7%,相比于原始算法效果在参数量、权重文件大小等方面均有较大提升。最后应用于研制的火车车厢自动摘钩机器人中,极大提高了自动摘钩的准确性和有效性。

    Abstract:

    It is of great significance to identify the coupling handle of the train coupler accurately for the work of the uncoupling robot. To solve this problem, a target recognition algorithm based on improved YOLOv5 train connection handle is proposed. Due to the few types of coupler connection handles in the production site and the complex production environment, it is difficult to ensure a rich picture data set. It is necessary to perform data expansion processing on the data set first, including a series of operations such as rotation, translation and brightness adjustment. Simulate the image information detected by cameras in different production environments. Then SAC module is used to replace C3 module in the backbone feature extraction network, and WIOU(wise intersection over union) function is introduced as a new bounding box loss function. it can enhance the feature extraction ability of the backbone network, improve the generalization ability and convergence rate of the model. Experimental results show that the target recognition rate of the improved YOLOv5 algorithm on the train connection handle reaches 97.7%, which is significantly improved compared with the original algorithm. It has obvious advantages in the number of parameters, the size of weight file, etc. It has a good detection effect on the train connection handle, and can be better deployed to the actual production.

    参考文献
    相似文献
    引证文献
引用本文

刘雷,王野,田长河,等. 基于YOLOV5的火车连接手柄识别方法及在自动摘钩机器的应用[J]. 科学技术与工程, , ():

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2023-11-21
  • 最后修改日期:2024-06-12
  • 录用日期:2024-06-24
  • 在线发布日期:
  • 出版日期:
×
亟待确认版面费归属稿件,敬请作者关注
《科学技术与工程》入选维普《中文科技期刊数据库》自然科学类期刊月度下载排行榜TOP10