基于数字孪生的印刷电路板缺陷检测算法优化
Optimization of Defect Detection Algorithm for Printed Circuit Board Based on Digital Twins
迟盛元 1白岩 1孟祥民 1杨海龙 1李洪洲1
作者信息
- 1. 北华大学机械工程学院,吉林吉林 132013
- 折叠
摘要
针对传统生产线上缺陷检测过程中实时监控性能差、检测精度不足等问题,提出了基于数字孪生的缺陷检测算法.根据检测设备的运行特点和实时监控的要求,提出了数字孪生的基础框架.以微型电子产品装配产线中的视觉检测机器人为例,基于数字孪生框架,构建了机器人的数字孪生系统.利用深度学习的知识,对缺陷检测算法进行优化.实验结果表明,改进后的算法准确率达到了 96.5%,提高了视觉检测的智能化程度,为机器人智能化管理提供了新的思路.
Abstract
In view of the poor real-time monitoring performance and insufficient detection accuracy in the process of defect detection on traditional production lines,a digital twins-based algorithm for defect detection is proposed.Firstly,based on the operating characteristics of the detection equipment and the requirement of real-time monitoring,a basic framework for digital twins is proposed.Secondly,taking the visual inspection robot in the assembly line of microelectronic products as an example,a digital twins system for the robot is constructed based on the digital twins framework.Finally,leveraging knowledge in deep learning,the defect detection algorithm is optimized.Experimental results demonstrate that the improved algorithm achieved an accuracy rate of 96.5%,enhancing the intelli-gence level of visual detection and providing new insights for the intelligent management of robots.
关键词
机器人/数字孪生/深度学习/缺陷检测Key words
robot/digital twins/deep learning/defect detection引用本文复制引用
基金项目
北华大学博士基金项目()
吉林省高教科研重点课题(JGJX2023C49)
吉林省科技发展计划项目(20210203109SF)
吉林省教育厅科学技术研究项目(JJKH20210035KJ)
出版年
2024
国家科技期刊平台
NSTL
万方数据