面向铁水温度的高炉异常数据检测及修补  

Furnace Temperature Modelling and Data Processing for Blast Furnace Object Oriented Hot Metal Temperature

作　　者：赵哲[1] 张勇[1] 于楠楠[1] 崔桂梅[1] 

机构地区：[1]内蒙古科技大学信息工程学院,包头014010

出　　处：《自动化与仪表》2015年第2期63-67,共5页Automation & Instrumentation

基　　金：国家自然科学基金资助项目(61164018);内蒙古自治区研究生教育创新计划资助项目(S20141012702);内蒙古自然科学基金资助项目(2014MS0612);内蒙科技计划资助项目(41402060423)

摘　　要：铁水温度是表征高炉炉温的重要参数,受高炉数据采集系统影响,存在数据缺失及异常等问题,给高炉建模及操作优化带来较大的困难。针对常规异常数据检测及修补方法存在的误检测及修补精度低等问题,该文研究高炉异常数据检测及修补新方法。首先,从高炉操作的角度介绍现场数据采集的特点;然后,针对现场采集的数据存在的具体问题,用常规方法对异常数据进行检测、处理及修补,并说明存在的不足;进而提出以出铁批次为时间序列的多尺度的铁水温度异常数据检测及修补新方法。最后,应用某钢厂高炉数据做模型试验,结果表明该方法具有良好的有效性。Hot metal temperature was an important characterize for furnace temperature representation,the outliers and missing values in collected data sets bring greater difficulties to blast furnace modelling and operation optimization. This work will study the outliers and missing values new detection and recovery techniques due to the problems of error detection and low accuracy recovery by traditional detection and recovery methods. This paper begins with the blast furnace data aquisition background introduction from blast furnace operation angle. Then,according to the spe- cific problems in data collected system,this work uses conventional methods to process and repair the outliers and missing values,and points out the problem. Furthermore,this paper proposes a new detection and recovery methods for hot metal temperature outliers and missing values based on multi-sacle technique,which takes the tapping num- ber series as the time axis. Finally,a steel mill blast furnace data set was given to test the proposed methods,the results show that the new methods have better performance than traditional methods.

关 键 词：炉温 铁水温度 建模 数据处理 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]