基于大数据的高炉炼铁数据治理及挖掘的研究  

Research on data management and mining of blast furnace ironmaking based on big data

作　　者：何家雷 唐珏 储满生 石泉 张振 HE Jialei;TANG Jue;CHU Mansheng;SHI Quan;ZHANG Zhen(School of Metallurgy,Northeastern University,Shenyang 110819,Liaoning,China)

机构地区：[1]东北大学冶金学院,辽宁沈阳110819

出　　处：《钢铁研究学报》2024年第7期844-854,共11页Journal of Iron and Steel Research

基　　金：国家自然科学基金资助项目(52274326);中国宝武低碳冶金创新基金资助项目(BWLCF202109)。

摘　　要：针对目前高炉炼铁数据的问题和特点,提出了一种关于高炉炼铁数据预处理及分析挖掘的研究方案。以整合的多源异构高炉数据为基础,结合炼铁经验和大数据技术对数据进行预处理。采用重采样处理高炉数据频次差异问题,对高炉不同类型的缺失数据进行综合填补,利用箱型图识别异常值并对异常值进行填补,完成了数据的高质量化。针对高炉参数影响规律分析发现,热风温度改变2~3 h后,焦比和铁水[Si]含量随之发生变化。通过相关性分析得到高炉参数间的关联度,采用K-means融合Apriori算法计算高炉参数间的关联规则,完成模糊关联的数值精准化。基于高炉生产数据并以指标参数焦比为例,发现原燃料参数中焦炭CRI(%)数值范围在(12.13,16.55]、烧结矿MgO质量分数(%)数值范围在(2.15,2.33]、焦炭M_t(%)数值范围在(2.85,7.96]时,有助于高炉焦比的降低。A research scheme on preprocessing and analysis mining of blast furnace ironmaking data was proposed in response to the current problems and characteristics of blast furnace ironmaking data.Based on integrated multi-source heterogeneous blast furnace data,combined with ironmaking experience and big data technology,the data was preprocessed.By using resampling to address the issue of frequency differences in blast furnace data,and the missing data of different types were comprehensively filled,and the outliers were identified and flled by the Box-plot to complete the high quality quantification of data.According to the analysis of the influence law of blast furnace parameters,it is found that after changing the hot air temperature for 2-3 h,the coke ratio and[Sij content in molten iron also change.The correlation degree among blast furnace parameters was obtained through correlation analysis,and the association rules among blast furnace parameters were calculated by K-means and Apriori algorithm to complete the numerical precision of fuzzy association.Based on the production data of blast furnace and taking the coke ratio as an example,it was found that the range of CRI(%)values for coke in the raw fuel parameters is(12.13,16.55],the range of MgO content for sintered ore is(2.15,2.33],and the range of Mt values for coke is(2.85,7.96),which helps to reduce the coke ratio of the blast furnace.

关 键 词：高炉炼铁 数据预处理 时滞性分析 相关性分析 关联规则挖掘 

分 类 号：TF54[冶金工程—钢铁冶金]