• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2019, Vol. 55 ›› Issue (17): 1-7.doi: 10.3901/JME.2019.17.001

• 特邀专栏:焊接机器人 •    

基于视觉信息融合的熔丝成形制造路径在线识别

常树鹤1, 张昊宇1, 许海鹰2, 洪宇翔1, 王力1, 都东1   

  1. 1. 清华大学机械工程系 北京 100084;
    2. 北京航空制造工程研究院 北京 100024
  • 收稿日期:2018-09-26 修回日期:2019-03-06 发布日期:2020-01-07
  • 通讯作者: 都东(通信作者),男,1962年出生,博士,长聘教授。主要研究方向为机器人与智能成形制造。E-mail:dudong@tsinghua.edu.cn
  • 作者简介:常树鹤,男,1993年出生,博士研究生。主要研究方向为电子束熔丝沉积增材制造过程智能控制。E-mail:changsh15@mails.tsinghua.edu.cn
  • 基金资助:
    国家自然科学基金(51875309)、国家工信部基金(xx-079)和国家重点研发计划(2017YFB1103100)资助项目。

Online Path Recognition for Additive Manufacturing and Arc Welding with Filler Wire Based on Multi-vision Information Fusion

CHANG Shuhe1, ZHANG Haoyu1, XU Haiying2, HONG Yuxiang1, WANG Li1, DU Dong1   

  1. 1. Department of Mechanical Engineering, Tsinghua University, Beijing 100084;
    2. Beijing Aeronautical Manufacturing Technology Research Institute, Beijing 100024
  • Received:2018-09-26 Revised:2019-03-06 Published:2020-01-07

摘要: 在能源装备制造领域大型构件多层多道电弧焊接和航空制造领域重要构件电子束熔丝沉积增材制造过程中,迫切需要路径在线识别及实时自动导引技术。然而在熔丝制造过程中,相邻焊道之间的几何差异很小,传统的结构光焊缝识别技术无法适应。为解决该问题,针对多层多道电弧焊和电子束熔丝沉积增材制造两种场景,分别设计基于双侧定向光影与结构光信息融合的视觉检测装置,并进行成形试验。试验结果表明,将双侧定向光源单独点亮时获得的焊缝灰度图像作差时,焊道边界附近呈现明显的"高灰度-低灰度-高灰度"的过渡区域。时域融合处理可以有效消除弧光、飞溅的干扰。提出的定向光影视觉信息与结构光传感信息实时融合处理方法,较之已有基于线结构光信息的焊道识别和自动跟踪系统在某些场合具有更强适应性,为保证成形质量和提高生产效率提供了新的技术途径。

关键词: 多层多道电弧焊接, 熔丝沉积增材制造, 路径在线识别, 视觉检测, 信息融合

Abstract: In the manufacturing process of multi-layer/multi-pass weld of large-scale components in the field of energy equipment and electron beam freeform fabrication in the field of aerospace manufacturing, there is an urgent need for online recognition and real-time automatic guidance technology. But the geometrical differences is very small between adjacent weld passes which makes it difficult to detect the weld position using traditional methods such as structured light. A method of real-time fusion processing of directional and structured light information is developed to solve this problem, which can be used in multi-layer/multi-pass weld and electron beam freeform fabrication. Corresponding visual detection devices are also designed. The result shows that there is a "High-Low-High" distribution of brightness between adjacent passes when using differential image. The act of fusing adjacent images greatly reduces the influence of the spatters and arc light. Compared with the existing weld path recognition and automatic tracking system based on structured light method, it has stronger adaptability to ensure the processing quality. This method also provides a new technical approach to improve the productivity.

Key words: multi-layer/multi pass arc welding, wire feed additive manufacturing, in-situ seam tracking, information fusion, visual detection

中图分类号: