铝电解槽寿命相关数据分析与可视化研究  

Study on Data Analysis and Visualization of Lifespan of Aluminum Electrolysis Cells

作　　者：李劼 施辉洪[1,2] 彭晨 陈灿 陈开斌 罗英涛[4] 张亚楠 王怀江 张红亮 LI Jie;SHI Huihong;PENG Chen;CHEN Can;CHEN Kaibin;LUO Yingtao;ZHANG Yanan;WANG Huaijiang;ZHANG Hongliang(School of Metallurgy and Environment,Central South University,Changsha 410083,China;National Engineering Research Center for Low-carbon Non-ferrous Metallurgy,Central South University,Changsha 410083,China;Microsoft(China)Co.,Ltd.,Wuxi Branch,Wuxi 214000,Jiangsu,China;Chinalco Zhengzhou Nonferrous Metals Research Institute Co.,Ltd.,Zhengzhou 450000,China)

机构地区：[1]中南大学冶金与环境学院,长沙410083 [2]中南大学低碳有色冶金国家工程研究中心,长沙410083 [3]微软(中国)有限公司无锡分公司,江苏无锡214000 [4]中铝郑州有色金属研究院有限公司,郑州450000

出　　处：《有色金属（冶炼部分）》2024年第11期138-149,共12页Nonferrous Metals（Extractive Metallurgy)

基　　金：国家自然科学基金资助项目(U2202253,62133016);云南省重大科技专项计划资助项目(202202AB080017);中南大学前沿交叉项目(2023QYJC007)。

摘　　要：随着工业信息化的深入发展,铝电解生产过程中产生的数据量日益庞大,传统的数据分析方法在应对复杂的工业环境时显得不足。提出了一套基于边缘端-云端实时感知系统的铝电解大数据分析方案,集成了在线监测、人工离线检测及记录数据。首先使用经典统计分析方法对铝电解槽寿命及工艺参数进行了初步研究,发现平均寿命为2064 d,运行时长超过2000 d的槽占58%。接着,应用探索性数据分析(EDA)方法对数据进行了缺失、噪声和特征缩放等预处理,并通过可视化方法揭示了工艺参数的分布特征及其相关性。最终,提取出关键变量,为铝电解槽寿命预测模型的建立提供了基础数据。With the deepening development of industrial informatization,the volume of data generated during aluminum electrolysis production has increased significantly,rendering traditional data analysis methods insufficient for addressing complex industrial environments.In this paper,a big data analysis framework for aluminum electrolysis based on an edge-cloud real-time sensing system was proposed,which integrates online monitoring,manual offline detection and recorded data.Firstly,the classical statistical analysis methods were employed to conduct an initial study on the lifespan and process parameters of aluminum electrolysis cells,revealing an average lifespan of 2064days,with 58%of the cells operating for over 2000days.Then,the Exploratory Data Analysis(EDA)method was applied to perform data preprocessing,including handling missing values,noise reduction,and feature scaling.The distribution characteristics and correlations of process parameters are visualized.Finally,the key variables were extracted,which provides foundational data for the establishment of the lifespan prediction model for aluminum electrolysis cells.

关 键 词：电解铝 槽寿命 探索性数据分析 分布式感知 

分 类 号：TF821[冶金工程—有色金属冶金]