摘要
掘锚机滚筒在遇到坚硬岩石时,其截齿座会发生磨损甚至整体断裂。掘锚机滚筒的截齿座发生孔口磨损变形后,截齿在使用过程中不能再自由转动,截齿会发生单侧面截割情况,这样会导致截齿局部的熔覆合金或合金头快速磨损而损坏,降低使用寿命。损坏的截齿座在修复过程中,抓取截齿座时需要准确找到截齿座圆孔中心,使得机器人手臂可以准确抓取,但由于损坏截齿座的形状复杂,圆心识别难度大,识别精度低。因此开发了基于Segment Anything深度学习模型的中心识别算法,实现了复杂形状截齿座的高精度中心识别及分类。开发了平面拟合算法及坐标转换算法,实现了精确的圆孔中心识别定位。识别精度达到了1.7 mm,抓取成功率100%,实现了准确的截齿座中心定位抓取。
When the anchor excavator drum encounters hard rock,its pick seat will wear out or even break wholly.After the orifice of the pick seat of the excavator drum is worn and deformed,the pick cannot rotate freely during use,and the pick will be cut on one side,which will lead to the rapid wear and damage of the cladding alloy or alloy head of the pick,and reduce the service life.In the repair process of the damaged pick seat,the center of the circular hole of the pick seat needs to be accurately found when grabbing the pick seat,so that the robot arm can accurately grasp,but due to the complex shape of the damaged pick seat,the center of the circle is difficult to identify,and the recognition accuracy is low.Therefore,a center recognition algorithm based on Segment Anything deep learning model was developed to achieve high-precision center recognition and classification of complex shape pick seats.The plane fitting algorithm and coordinate conversion algorithm were developed,and the accurate identification and positioning of the circular hole center were realized.The recognition accuracy has reached 1.7 mm,the grasping success rate is 100%,and achieves the accurate positioning and grasping of the center of the pick seat.
作者
刘强
Liu Qiang(Equipment Maintenance Center,CHN Energy Shendong Coal Group,Yulin 719000,China)
出处
《煤矿机械》
2025年第3期64-67,共4页
Coal Mine Machinery
关键词
深度学习
截齿座
中心识别
目标定位
deep learning
pick seat
center identification
target location
