基于机器视觉的镍硫金属铁含量检测系统  

Machine vision based detection system for nickel-sulfur metal iron content

作　　者：刘仲民[1] 崔勇 吴海宝 苏鹏飞 LIU Zhongmin;CUI Yong;WU Haibao;SU Pengfei(College of Electrical and Information Engineering,Lanzhou University of Technology,Lanzhou 730050,China;Jinchuan Group Information and Automation Engineering Co.,Ltd.,Jinchang 737100,China)

机构地区：[1]兰州理工大学电气工程与信息工程学院,甘肃兰州730050 [2]金川集团信息与自动化工程有限公司,甘肃金昌737100

出　　处：《冶金自动化》2023年第4期100-108,共9页Metallurgical Industry Automation

基　　金：甘肃省工业过程先进控制重点实验室开放基金项目(2022KX10)。

摘　　要：针对镍转炉吹炼过程中环境恶劣、操作人员判断出炉时间有差异等问题,应用机器视觉技术研发一套镍硫金属铁含量在线检测系统。首先,使用工业相机采集镍硫金属样本的断面图像并进行图像分割与预处理。然后,利用图像纹理信息中的对比度和熵值进行样本选择。最后,提取图像颜色特征信息,利用灰度值特征和实测铁含量数据建立基于线性回归的一次模型与二次模型并进行对比,用评价指标进行评估。试验结果表明,铁含量检测二次模型拟合程度较好,准确率较高,且检测精度满足金属组分含量检测中最大相对误差小于5%的精度要求。In view of the harsh environment in the nickel converter blowing process and the difference in the operator's judgment of the furnace discharge time,machine vision technology was applied to develop a set of online nickel-sulfur metal iron content online detection system.Firstly,a cross-sectional image of the nickel-sulfur metal sample was acquired using an industrial camera and image segmentation and pre-processing were performed.Then,sample selection was performed using the contrast and entropy values from the image texture information.Finally,the image color feature information was extracted,and the primary model and the secondary model based on linear regression was established and compared using the gray value feature and the measured iron content data,and evaluated with the evaluation index.The experimental results show that the secondary model for iron content detection fits better and has higher accuracy,and the detection accuracy meets the accuracy requirement of maximum relative error less than 5%in metal component content detection.

关 键 词：镍转炉吹炼 铁含量检测 机器视觉 图像处理 线性回归 

分 类 号：TF815[冶金工程—有色金属冶金] TP391.41[自动化与计算机技术—计算机应用技术]