第10期 赵晨熙等:COREX冷煤气成分预测的二步建模方法 ·1189· 料单所对应的生产数据建立模型进行预测的预测相 (李海峰,王臣,邹宗树,等.COREX喷煤模型及应用分析. 对误差小39.96%和1.88%,预测均方根误差小 过程工程学报,2009,9(增刊1):349) B]Liu X G,Zeng J S,Zhao M.Mathematical model and its hybrid 3.138和0.2. dynamic mechanism in intelligent control of ironmaking.ron 一倾测样本真实值 Steel Res Int,2007,14(1)7 16 一+·类别B预测值 4]Xu W R,Guo Y L.Wang C.Analysis of the factors affecting the --类别C预测值 14 未聚类预测值 fuel rate in the COREX process and improvement measures.Baost- eel Tech Res,2011,5(1):45 [5]Prachethan K P,Garg L M,Gupta S S.Modelling of Corex process for optimisation of operational parameters.fronmaking Steelmaking,2006,33(1):29 [6]Barman S C,Mrunmaya K P,Ranjan M.Mathematical model de- velopment of raceway parameters and their effects on Corex process.J Iron Steel Res Int,2011,18(5):20 10 15202530354045 7]Prachethan K P,Dasu A V R P,Ranjan M,et al.Influence of 样本综号 operational parameters on silicon in hot metal from Corex.fron- 图6不同方法预测得到的冷煤气C02含量结果比较 making Steelmaking.2008,35 (2)108 Fig.6 Comparison between the predicted results of cold gas CO []Jia GL Process Calculation and Study on Pulverized Coal produc- content by using different prediction methods tion in COREX Melter Gasifier [Dissertation ]Chongqing: Chongqing University,2007 (贾国利.COREX熔脸气化炉工艺计算及煤粉在炉内燃烧行 3结论 为研究[学位论文].重庆:重庆大学,2007) 本文提出了基于熵权模糊C均值聚类和偏最 9] Fang J,Wang X J,Shi Y,et al.Non Blast Furnace lronmaking 小二乘的COREX冷煤气成分含量预测方法,以宝 Process and Theory.2nd Ed.Beijing:Metallurgical Industry Press,2010 钢COREX-1"炉的实际生产数据为实例,对冷煤气 (方觉,王杏娟,石炎,等.非高炉炼铁工艺与理论.2版.北 C0,含量建立了预测模型.现场数据的实验结果表 京:治金工业出版社,2010) 明:基于熵权模糊C均值聚类和偏最小二乘回归方 [10]Shannon C E.A mathematical theory of communication.Bell Syst 法建立的COREX冷煤气CO,含量预测模型可以较 TehJ,1948,27:379 好地预测出实际的C02含量,为冷煤气成分含量的 1]Berdek JC.Pattern Recognition with Fuy Objectire Function Al- gorithms.Norwell:Kluwer Academic Publishers,1981 控制提供了有效的分析手段. [12]Kaufman L.Rousseeuw P J.Finding Groups in Data:an Intro- duction to Cluster Analysis.Wiley Online Library,1990 参考文献 [13]Wang H W,Wu Z B,Meng J.Partial least-squares regression linear and nonlinear methods.Beijing:National Defence Industry [Wu K,Zhang E H.Wan P,et al.Consideration of the theoretical Pres5,2006 combustion temperature formula before tuyere in melting gasifier at (王惠文,吴载斌,孟洁.偏最小二乘回归的线性与非线性 COREX process.J China Coal Soc,2010,35(10):1730 方法.北京:国防工业出版社,2006) (吴铿,张二华,万鹏,等.关于COREX流程熔融气化炉风口 41 Zhang JC,Wu C.Non-inear prediction model for coke thermal 前理论燃烧温度的思考.煤炭学报,2010,35(10):1730) properties based on partial least squares regression.J Cent South Li H F,Wang C,Zou Z S,et al.Coal injection model for Unir Sci Technol,2011,42(5):1406 COREX and its application analysis.Process Eng,2009,9 (张进春,吴超.基于偏最小二乘回归的焦炭热性质非线性 (Suppl 1):349 预测模型.中南大学学报:自然科学版,2011,42(5):1406)第 10 期 赵晨熙等: COREX 冷煤气成分预测的二步建模方法 料单所对应的生产数据建立模型进行预测的预测相 对误 差 小 39. 96% 和 1. 88% ,预 测 均 方 根 误 差 小 3. 138 和 0. 2. 图 6 不同方法预测得到的冷煤气 CO2含量结果比较 Fig. 6 Comparison between the predicted results of cold gas CO2 content by using different prediction methods 3 结论 本文提出了基于熵权模糊 C 均值聚类和偏最 小二乘的 COREX 冷煤气成分含量预测方法,以宝 钢 COREX--1# 炉的实际生产数据为实例,对冷煤气 CO2含量建立了预测模型. 现场数据的实验结果表 明: 基于熵权模糊 C 均值聚类和偏最小二乘回归方 法建立的 COREX 冷煤气 CO2含量预测模型可以较 好地预测出实际的 CO2含量,为冷煤气成分含量的 控制提供了有效的分析手段. 参 考 文 献 [1] Wu K,Zhang E H,Wan P,et al. Consideration of the theoretical combustion temperature formula before tuyere in melting gasifier at COREX process. J China Coal Soc,2010,35( 10) : 1730 ( 吴铿,张二华,万鹏,等. 关于 COREX 流程熔融气化炉风口 前理论燃烧温度的思考. 煤炭学报,2010,35( 10) : 1730) [2] Li H F,Wang C,Zou Z S,et al. Coal injection model for COREX and its application analysis. J Process Eng,2009,9 ( Suppl 1) : 349 ( 李海峰,王臣,邹宗树,等. COREX 喷煤模型及应用分析. 过程工程学报,2009,9( 增刊 1) : 349) [3] Liu X G,Zeng J S,Zhao M. Mathematical model and its hybrid dynamic mechanism in intelligent control of ironmaking. J Iron Steel Res Int,2007,14( 1) : 7 [4] Xu W R,Guo Y L,Wang C. Analysis of the factors affecting the fuel rate in the COREX process and improvement measures. Baosteel Tech Res,2011,5( 1) : 45 [5] Prachethan K P,Garg L M,Gupta S S. Modelling of Corex process for optimisation of operational parameters. Ironmaking Steelmaking,2006,33( 1) : 29 [6] Barman S C,Mrunmaya K P,Ranjan M. Mathematical model development of raceway parameters and their effects on Corex process. J Iron Steel Res Int,2011,18( 5) : 20 [7] Prachethan K P,Dasu A V R P,Ranjan M,et al. Influence of operational parameters on silicon in hot metal from Corex. Ironmaking Steelmaking,2008,35( 2) : 108 [8] Jia G L. Process Calculation and Study on Pulverized Coal production in COREX Melter Gasifier [Dissertation]. Chongqing: Chongqing University,2007 ( 贾国利. COREX 熔融气化炉工艺计算及煤粉在炉内燃烧行 为研究[学位论文]. 重庆: 重庆大学,2007) [9] Fang J,Wang X J,Shi Y,et al. Non Blast Furnace Ironmaking Process and Theory. 2nd Ed. Beijing: Metallurgical Industry Press,2010 ( 方觉,王杏娟,石炎,等. 非高炉炼铁工艺与理论. 2 版. 北 京: 冶金工业出版社,2010) [10] Shannon C E. A mathematical theory of communication. Bell Syst Tech J,1948,27: 379 [11] Bezdek J C. Pattern Recognition with Fuzzy Objective Function Algorithms. Norwell: Kluwer Academic Publishers,1981 [12] Kaufman L,Rousseeuw P J. Finding Groups in Data: an Introduction to Cluster Analysis. Wiley Online Library,1990 [13] Wang H W,Wu Z B,Meng J. Partial least-squares regression: linear and nonlinear methods. Beijing: National Defence Industry Press,2006 ( 王惠文,吴载斌,孟洁. 偏最小二乘回归的线性与非线性 方法. 北京: 国防工业出版社,2006) [14] Zhang J C,Wu C. Non-linear prediction model for coke thermal properties based on partial least squares regression. J Cent South Univ Sci Technol,2011,42( 5) : 1406 ( 张进春,吴超. 基于偏最小二乘回归的焦炭热性质非线性 预测模型. 中南大学学报: 自然科学版,2011,42( 5) : 1406) ·1189·