智能选矸机器人的研究与工程实践  

Research and engineering practice of intelligent gangueselection robot

作　　者：晋纪岩 赵珍超 JIN Jiyan;ZHAO Zhenchao(Xintun Mine,Jizhong Energy Fengfeng Group Co.,Ltd.,Handan 056200,Hebei,China)

机构地区：[1]冀中能源峰峰集团有限公司新屯矿,河北邯郸056200

出　　处：《矿山机械》2023年第11期43-48,共6页Mining & Processing Equipment

摘　　要：智能选矸机器人是基于机器视觉的智能煤矸识别,通过高速相机摄取煤和矸石图像,对煤块和矸石的物状、纹理、灰度、波长、反射光线等数据进行采集并标定入库,再经一级目标检测网络识别煤矸轮廓、二级网络进行煤矸初筛和精筛分类,采用神经网络技术深度学习,从而实现煤矸识别功能。识别后的煤矸数据坐标,经多组多轴仿生机械手模拟人工抓取和拨取动作,实现快速精准抓取矸石。针对当前智能选矸机器人存在的一些问题,提出了改进措施。The intelligent gangue selection robot could identify the coal gangue intelligently based on the machine vision.It took images of the coal and the gangue through high-speed cameras,collected and calibrated data such as physical shape,texture,gray scale,wavelength and reflected light of the coal and the gangue,and then identified the outline of the coal gangue through first-level target detection network,conducted preliminary screening and fine screening classification of the coal gangue through second-level network,and adopted neural network technology for deep learning.Thus the function of identifying coal waste could be realized.After identifying the data coordinates of the coal waste,multiple groups of multi-axis bionic mechanical arms simulated the manual grasping and pulling action to achieve fast and accurate grasping of gangue.In view of the problems existing in the current intelligent gauge selection robot,the improvement measures were put forward.

关 键 词：煤矸分离 智能选矸机器人 AI视频识别 神经网络学习 仿生机械手 

分 类 号：TD94[矿业工程—选矿]