摘要
印刷电路板(PCB)检测方法对于确保产品正常工作至关重要。该文针对传统的人工检测方法易存在漏检、误检等问题,采用深度学习方法对PCB缺陷进行检测,并搭建了基于ZYNQ的硬件实现平台;采用软硬件协同设计方法,使用FPGA对算法进行了硬件加速,其中包括采用了YOLOv3-SPP网络模型,并对该结构进行了优化,使其适用于ZYNQ端的部署。在搭建硬件平台时,首先通过Vivado配置硬件基本信息,然后使用PetaLinux创建Linux系统,在Vitis中调用该系统并添加DPU IP核,最后在ZYNQ的PS端采用多线程思想编写Python程序,实现PCB缺陷的检测。实验结果显示,该系统对各类型PCB缺陷的检测精度均在0.95以上,检测精度平均值(mAP)为0.97。
The detection method of printed circuit board(PCB)is crucial to ensure the normal operation of the product.In view of the problems that traditional manual detection methods are prone to miss detection and false detection,this paper uses the deep learning method to detect PCB defects,and builds a hardware implementation platform based on ZYNQ.It uses the software and hardware co-design method to accelerate the algorithm using FPGA.Among them,YOLOv3-SPP network model is adopted,and the structure is optimized to make it suitable for the deployment of ZYNQ terminal.When building the hardware platform,first configure the basic hardware information through Vivado,then use PetaLinux to create a Linux system,call the system in Vitis and add the DPU IP core,and finally write Python programs on the PS side of ZYNQ with the idea of multithreading to achieve PCB defect detection.The experimental results show that the detection accuracy of the system for various types of PCB defects is above 0.95,and the average detection accuracy(mAP)is 0.97.
作者
张丽丽
陈真
刘雨轩
蔡健楠
ZHANG Lili;CHEN Zhen;LIU Yuxuan;CAI Jiannan(College of Electronical and Information Engineering,Shenyang Aerospace University,Shenyang 110136,China)
出处
《实验技术与管理》
CAS
北大核心
2023年第4期96-102,共7页
Experimental Technology and Management
基金
国家自然科学基金资助项目(61671310)
辽宁省教育厅资助项目(LJKZ0174)
辽宁省教改资助项目(辽教办[2021]254号)。
作者简介
张丽丽(1979—)女,黑龙江讷河,博士,副教授,系主任,主要研究方向为FPGA系统设计与深度学习,20052727@sau.edu.cn。
