神经网络在自动扒渣控制优化中的应用  

Application of Neural Networks in Optimization of Automatic Slag Skimming Control

作　　者：宋扬 王渤涵 冯岭[1] 张世凯[1] 苏睿聪 SONG Yang;WANG Bohan;FENG Ling;ZHANG Shikai;SU Ruicong(Beijing Shougang Automation Information Technology Co.,Ltd.,Beijing 100041,China)

机构地区：[1]北京首钢自动化信息技术有限公司,北京100041

出　　处：《自动化应用》2024年第2期16-19,共4页Automation Application

摘　　要：某钢厂计划改造KR脱硫站扒渣位,前期进行了图像分析的软硬改造,但受限于成像位置,自动控制效果不佳。通过引入神经网络预测倾翻终点的位置,转换像素坐标,提高了自动倾翻终点的准确度,提升了自动扒渣控制扒渣臂的精度,使整个自动扒渣流程更精确流畅。A steel mill plans to renovate the KR desulfurization station slag skimming position,the previous soft and hard transformation of image analysis,but limited by the imaging position,the aucomatic control effect is not good.Through the introduction of neural networks to predict the position of the tipping end point,the pixel coordinates are converted to improve the accuracy of the automatic tipping end point,enhance the accuracy of the automatic slag skimming control slag raking arm,and make the whole automatic slag skimming process more accurate and smooth.

关 键 词：神经网络 自动化 扒渣 像素坐标转换 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]