摘要
本文采用YOLOv3算法计算轮毂螺丝检测框的位置,并通过建立坐标系来分析轮毂和螺丝畸变方向的一致性,将轮毂误差代入螺丝角度以消除螺丝检测框相对于轮毂的误差,从而计算轮毂的螺丝角度,以解决畸变图像下车轮故障检测困难的问题。结果表明:YOLOv3对链箅机螺丝的位置检测具有可行性,能计算出螺丝角度,其计算误差在±10°以内。
YOLOv3 algorithm is adopted to calculate the position of the wheel hub and screw detection frame,and the consistency of the wheel hub and screw distortion direction is analyzed by establishing a coordinate system,and then the wheel hub error is substituted into the screw angle to eliminate the error of the screw detection frame relative to the wheel hub and the wheel screw angle is calculated for solving the problem of difficult detection of wheel faults under distorted images.The results show that YOLOv3 is feasible for detecting the position of screws of chain grate and can calculate the screw angle with a calculation error of±10°or less.
作者
房良睿
王月明
李真如
刘浩楠
吴永刚
FANG Liangrui;WANG Yueming;LI Zhenru;LIU Haonan;WU Yonggang(Information Engineering School,Inner Mongolia University of Science and Technology,Baotou 014010,Inner Mongolia,China;Baotou Ironmaking Plant,Baotou 014010,Inner Mongolia,China)
出处
《烧结球团》
北大核心
2021年第1期76-82,共7页
Sintering and Pelletizing
基金
国家自然科学基金资助项目(61962046)
内蒙古自然基金资助项目(2020MS06008)。
