摘要
为了采集多类型工况的图像,扩大神经网络的学习样本。改变光源的高度能显著影响图像的差异,为此对影响光源位移准确度的步进电机和驱动结构进行误差分析并提出对应的优化补偿方法。经由对四台步进电机所需负载分析研究,建立负载模型,优化负载驱动电路。针对传动结构中轴心定位误差以及丝杆应力的干扰分别设计优化方案并完成对整体方案的验证。实验表明优化方法能有效地减小装置误差,四组传动丝杆的定位精度分别提升了88.896%、98.336%、45.081%、31.479%,优化效果明显,为图像采集提供位移精度保证。
In order to acquire images of multiple types of working conditions,the learning sample of the neural network is expand-ed.Changing the height of the light source can significantly affect the differences in the images.For this reason,an error analysis of the stepper motors and drive structure,which affect the accuracy of the light source displacement,was carried out and a corre-sponding optimization compensation method was proposed.By analyzing the loads required by the four stepper motors,a load model is established and the load drive circuit is optimized.Optimization solutions are designed and validated for the interference of axial positioning errors and filament stresses in the drive structure.The experiments show that the optimization method can effectively re-duce the device error,and the positioning accuracy of the four sets of drive filaments is improved by 88.896%,98.336%,45.081%,and 31.479%respectively.The optimization effect is obvious and provides a displacement accuracy guarantee for image acquisition.
作者
甘勇
于江豪
曾勃乔
饶承剑
GAN Yong;YU Jiang-hao;ZENG Bo-qiao;RAO Cheng-jian(College of Electrical and Mechanical Engineering,Guilin University of Electronic Science and Technology,Guangxi Guilin 541004,China)
出处
《机械设计与制造》
北大核心
2024年第3期27-30,34,共5页
Machinery Design & Manufacture
基金
国家自然科学基金—基于网格片层体积测量的机械零件无损测量与重构方法研究(51665008)。
关键词
机器视觉
负载补偿
图像采集
误差分析
Machine Vision
Load Compensation
Image Acquisition
Error Analysis
