摘要
转炉冶炼过程中,对熔池中钢水的碳含量进行精确预测是终点预报技术的核心环节。然而,如何通过解析炉口火焰特性与终点碳含量之间的深层联系,以实现更精准的预测,仍是当前面临的挑战。针对火焰纹理在不同碳含量下呈现不规则性且具有高度相似性的问题,提出了一种多方向加权复杂网络的彩色纹理特征提取模型。该模型利用滑动窗口选出了中心点的邻域顶点以构建邻域关系,利用顶点对之间的色彩距离构建了加权边,从而构建出细致彩色纹理和区域彩色纹理复杂网络,最后用网络顶点度分布特征量化网络特性得到彩色纹理特征描述符,与提取的色彩特征结合得到火焰特征描述符,通过回归模型预测终点碳含量。通过对转炉炼钢实际生产数据的实验研究验证了所提方法的有效性。
In the process of Basic Oxygen Furnace(BOF)smelting,accurate prediction of the carbon content in molten steel within the bath represents a pivotal aspect of endpoint prediction technology.However,deciphering the profound correlation between the characteristics of the furnace mouth flame and the final carbon content to achieve more precise predictions remains a prevailing challenge.To address the issue of flame texture irregularity and high similarity under varying carbon contents,a multi-directional weighted complex network(MDWCN)model is established for extracting color texture features.This model employs a sliding window to select neighboring vertices around a central point,thereby establishing neighborhood relationships.Weighted edges are constructed using color distances between vertex pairs,leading to the formation of detailed color texture and regional color texture complex networks.Ultimately,the network's vertex degree distribution characteristics are utilized to quantify network properties,yielding a color texture feature descriptor.This descriptor,combined with extracted color features,forms a flame feature descriptor,enabling the prediction of final carbon content through a regression model.The effectiveness of the proposed method has been validated through experimental investigations using actual production data from converter steelmaking.
作者
刘建勋
刘辉
LIU Jianxun;LIU Hui(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,Yunnan,China)
出处
《钢铁研究学报》
北大核心
2025年第2期151-163,共13页
Journal of Iron and Steel Research
基金
国家自然科学基金资助项目(62263016)
云南省高校服务重点产业科技资助项目(FWCY-QYCT2024003)
云南省"兴滇英才支持计划"资助项目
云南省应用基础研究基金资助项目(202401AT070375)
云南省科技厅面上资助项目(202001AT070038)。
关键词
转炉炼钢
碳含量预测
复杂网络
图像识别
彩色纹理特征分析
BOF steelmaking
endpoint carbon content prediction
complex network
image recognition
color texture feature analysis
