基于大数据与人工智能的电磁冶金数字化系统  

ELECTROMAGNETIC METALLURGY DIGITAL SYSTEM BASED ON BIG DATA AND ARTIFICIAL INTELLIGENCE

作　　者：刘勇 廉功豪 范建通 宋平 刘晓明 刘泽熠 王强 Liu Yong;Lian Gonghao;Fan Jiantong;Song Ping;Liu Xiaoming;Liu Zeyi;Wang Qiang(Technology Management Department of HBIS Group Co.,Ltd.,Shijiazhuang 050000,Hebei;State Key Laboratory of Rolling and Automation,Northeastern University,Shenyang 110819,Liaoning;Steel Plant of Liaoning Yingkou Shigang Jingcheng Equipment Technology Co.,Ltd.,Yingkou 115000,Liaoning;State Key Laboratory of Rolling Technology and Continuous Rolling Automation,Northeastern University,Shenyang 110819,Liaoning)

机构地区：[1]河钢集团股份公司科技管理部,河北石家庄050000 [2]东北大学材料电磁过程研究教育部重点实验室,辽宁沈阳110819 [3]辽宁营口石钢京诚装备技术有限公司炼钢厂,辽宁营口115000 [4]东北大学轧制技术及连轧自动化国家重点实验室,辽宁沈阳110819

出　　处：《河北冶金》2024年第11期13-18,60,共7页Hebei Metallurgy

基　　金：辽宁省高等学校创新团队支持计划(LT2017011)。

摘　　要：电磁冶金与连铸技术工艺变量的优化是企业提高连铸坯质量的重要手段。然而,企业对连铸过程的机理解释以及工艺变量的调节方法严重滞后于生产。因此,提出了基于数值仿真、计算机视觉以及大数据挖掘等多种技术的融合实现连铸过程多物理场的可视化、连铸坯质量自动识别与智能检测以及电磁冶金技术与连铸工艺变量的耦合优化。电磁冶金数字化系统初步具备对结晶器内的流场、温度场、液相体积分数、溶质场、夹杂物分布和凝固坯壳表面应力等多物理场的可视化以及铸坯中心缺陷的数字化识别能力。其中,使用Unet模型对于中心缺陷的识别效果可达90%以上,使用二值化的方式可以有效评估中心疏松的级别。可以预测,电磁冶金数字化系统对于推进钢铁企业智能化发展具有重要的意义。The optimization of process variables in electromagnetic metallurgy and continuous casting technology is an important means for enterprises to improve the quality of continuous casting billets.However,the explanation of the mechanism of continuous casting process and the adjustment methods of process variables by enterprises are seriously lagging behind production.Therefore,this article proposes the integration of multiple technologies such as numerical simulation,computer vision,and big data mining to achieve visualization of multiple physical fields in the continuous casting process,automatic identification and intelligent detection of continuous casting billet quality,and coupling optimization of electromagnetic metallurgy technology and continuous casting process variables.The electromagnetic metallurgy digital system has the preliminary ability to visualize multiple physical fields such as flow field,temperature field,liquid phase volume fraction,solute field,inclusion distribution,and surface stress of solidified billet shell in the crystallizer,as well as to digitally identify central defects in the casting billet.Among them,the recognition effect of the Unet model on central defects can reach over 90%,and the use of binarization can effectively evaluate the level of central looseness.It can be predicted that the digital system of electromagnetic metallurgy is of great significance for promoting the intelligent development of steel enterprises.

关 键 词：大数据 人工智能 电磁冶金 物理场可视化 数据挖掘 智能检测 数字化系统 

分 类 号：TF777[冶金工程—钢铁冶金]