摘要
随着物联网、工业4.0、人工智能时代的到来,智能化生产已开始逐渐推广。在总装生产线中,对于电动工具使用频率与使用正确与否的智能、实时检测方案仍然是短板。以生产线中大范围使用的电动螺丝刀为切入点,基于百度深度学习平台,利用其中PaddleSeg的图像分割技术和FSM有限状态机相结合,设计了一种能实时检测电动螺丝刀工作状态的算法,并设计了配套的智能检测硬件系统,训练分割模型和算法模型,实现了对电动螺丝刀工作状态的识别,并判断螺丝刀是否成功将螺丝嵌入器件,解决了当下工业生产中的难题。
With the advent of the Internet of things,Industry 4.0 and artificial intelligence era,intelligent factories become possible and necessary.Intelligent production has been gradually promoted,but in the final assembly line,the intelligent and real-time detection scheme for the use frequency and correctness of electric tools is still a short board.Based on Baidu deep learning platform,this paper designs a real-time algorithm to detect the working state of Electric Screwdrivers by combining PaddleSeg’s image segmentation technology with FSM finite state machine,and designs a matching intelligent detection hardware system,training segmentation model and algorithm model.Thus,it realizes the identification of the working state of the electric screw driver,and judges whether the screw is successfully driven,and solves the actual pain point in the current industrial production.
作者
崔博森
曾庆宇
窦蓉蓉
CUI Bosen;ZENG Qingyu;DOU Rongrong(School of Electronic Science and Engineering,Nanjing University,Nanjing 210023,China)
出处
《实验室研究与探索》
CAS
北大核心
2021年第9期46-51,65,共7页
Research and Exploration In Laboratory
基金
大学生创新创业训练计划(202010284117x)
国家级大学生创新创业训练计划(G201910284125)
教育部产学合作协同育人项目(201801155013)
江苏省教育厅高教技术研究会高校教育信息化研究课题(2019JSETK008)。
关键词
电动螺丝刀
机器视觉
深度学习
有限状态机
electric screwdriver
machine vision
deep learning
finite state machine(FSM)