铸坯质量的多元统计分析及决策树规则研究
|
摘要
铸坯质量是连铸生产中的最主要的问题之一。铸坯质量缺陷众多,且质量缺陷的影响因素难以把握。铸坯表面和内部的缺陷影响了生产效率,严重时还可能延长铸机停机时间,从而造成生产损失;同时连铸生产中采集到的各种工艺参数值等海量数据中可能隐藏了某些与铸坯质量有关的各种数据和信息,需要进行分析。因此,研究铸坯质量影响因素,找出其中的质量缺陷形成规律,并且进行铸坯质量的统计分析和数据挖掘,从而控制连铸生产工艺有重要的意义。
本文采用多元统计分析对生产过程中产生的数据进行分析,建立了多元分析中的主成分分析数学模型和Logistic回归分析数学模型;对生产数据的统计分析获得了对铸坯质量影响最大的因素,是进行数据挖掘进行规则构建的重要依据,并对建立的分析模型进行了验证;应用数据挖掘技术,对海量数据进行数据挖掘和分类,运用数据挖掘中决策树的算法建立数据挖掘中的数据分类模型,同时建立数据仓库,通过数据仓库中的数据文件建立数据映射。根据多元统计分析的结果建立数据仓库中的数据挖掘模型,对生产过程中采集的数据进行数据挖掘和规则构建,得到铸坯质量和生产工艺参数互相影响的决策树规则。
文中多元统计采用SPSS统计分析软件进行,数据挖掘采用WEKA软件和ORACLE数据库;利用多元分析结果中对铸坯质量影响比较大的因素作为数据挖掘模型中的维度,进行数据挖掘和分类,构建决策树规则。验证了多元统计分析数学模型和数据仓库数据模型,数据挖掘模型的有效性。研究中对铸坯质量影响比较大的因素进行提取,利用影响因素进行数据挖掘生成决策树规则,并提出了提高铸坯质量的方法。
Slab quality is one of the most important issues about casting production. Casting production has very defects, and the factor of these defects is difficult to grasp. Slab surface and internal defects affecting the productivity, and seriously the machine may be extended downtime, which resulting in production loss; at the same time slab quality-related data and information which needed for analysis may be hidden in the mass data which collected in the production such as the various parameters. Therefore, studying the factors of affect the quality of the slab, identify the law of the formation of defects and do statistical analysis and data mine about the slab quality has important significance to the continuous casting technology.
In this paper, multivariate statistical analysis of the production process of data analysis, the multivariate analysis of principal component analysis of mathematical models and mathematical models Logistic regression analysis; production data for statistical analysis to obtain the greatest impact on the quality of the slab, Data mining is carried out to build an important basis for the rules and the establishment of the model are verified; application of data mining, data on the mass of data mining and classification, the use of data mining in the decision tree algorithm to build data mining of data classification Model, at the same time to set up a data warehouse, data warehouse of data files to set up a data mapping. According to the results of multivariate statistical analysis to set up a data warehouse of data mining models, the production process of collecting data for data mining and construction of the rules, get the quality of casting and production process parameters affecting each of the decision tree rules.
SPSS statistical analysis software used for multivariate statistical analysis to get the result of statistical analysis of data, application software WEKA and ORACLE database, and the use of multivariate analysis of the results of the slab greatest impact on the quality of the data mining model as a factor in the third dimension, data mining and the classification of the rules of building a decision tree. Verify the efficiency of mathematical models, data warehouse data model and data mining model. On the production of slab greatest impact on the quality of the extracted factors, factors that affect the use of data mining and decision tree classification generated rules, and to improve the quality of the casting tools and solutions.
本文采用多元统计分析对生产过程中产生的数据进行分析,建立了多元分析中的主成分分析数学模型和Logistic回归分析数学模型;对生产数据的统计分析获得了对铸坯质量影响最大的因素,是进行数据挖掘进行规则构建的重要依据,并对建立的分析模型进行了验证;应用数据挖掘技术,对海量数据进行数据挖掘和分类,运用数据挖掘中决策树的算法建立数据挖掘中的数据分类模型,同时建立数据仓库,通过数据仓库中的数据文件建立数据映射。根据多元统计分析的结果建立数据仓库中的数据挖掘模型,对生产过程中采集的数据进行数据挖掘和规则构建,得到铸坯质量和生产工艺参数互相影响的决策树规则。
文中多元统计采用SPSS统计分析软件进行,数据挖掘采用WEKA软件和ORACLE数据库;利用多元分析结果中对铸坯质量影响比较大的因素作为数据挖掘模型中的维度,进行数据挖掘和分类,构建决策树规则。验证了多元统计分析数学模型和数据仓库数据模型,数据挖掘模型的有效性。研究中对铸坯质量影响比较大的因素进行提取,利用影响因素进行数据挖掘生成决策树规则,并提出了提高铸坯质量的方法。
Slab quality is one of the most important issues about casting production. Casting production has very defects, and the factor of these defects is difficult to grasp. Slab surface and internal defects affecting the productivity, and seriously the machine may be extended downtime, which resulting in production loss; at the same time slab quality-related data and information which needed for analysis may be hidden in the mass data which collected in the production such as the various parameters. Therefore, studying the factors of affect the quality of the slab, identify the law of the formation of defects and do statistical analysis and data mine about the slab quality has important significance to the continuous casting technology.
In this paper, multivariate statistical analysis of the production process of data analysis, the multivariate analysis of principal component analysis of mathematical models and mathematical models Logistic regression analysis; production data for statistical analysis to obtain the greatest impact on the quality of the slab, Data mining is carried out to build an important basis for the rules and the establishment of the model are verified; application of data mining, data on the mass of data mining and classification, the use of data mining in the decision tree algorithm to build data mining of data classification Model, at the same time to set up a data warehouse, data warehouse of data files to set up a data mapping. According to the results of multivariate statistical analysis to set up a data warehouse of data mining models, the production process of collecting data for data mining and construction of the rules, get the quality of casting and production process parameters affecting each of the decision tree rules.
SPSS statistical analysis software used for multivariate statistical analysis to get the result of statistical analysis of data, application software WEKA and ORACLE database, and the use of multivariate analysis of the results of the slab greatest impact on the quality of the data mining model as a factor in the third dimension, data mining and the classification of the rules of building a decision tree. Verify the efficiency of mathematical models, data warehouse data model and data mining model. On the production of slab greatest impact on the quality of the extracted factors, factors that affect the use of data mining and decision tree classification generated rules, and to improve the quality of the casting tools and solutions.
引文
[1]吴国富,项静恬.多个变量分类和综合的多元分析方法[J].数理统计与管理,2006,14(6):5
[2]王立新.孢粉数据的回归方法[J].数理统计与管理,1995,14(4):2
[3]洪新杰.青霉素质量管理的统计方法[J].数理统计与管理,2005,14(4):2
[4]奚介荣,龚光荣,知识型制造企业与知识管理[J].机电一体化,2006(2):8-11
[5]王学仁,王松桂编译.实用统计分析[M].上海:上海科学技术出版杜,1999
[6]王学仁等.实用多元统计分析[M].上海:上海科学技术出版社,2000
[7]秘晓,鲍玉斌,常钊等.工业数据仓库设计方法及其在质量分析中的应用[J].控制与决策,2007(2):27-34
[8] Haald Peters(德)等.采用基于数据方法各工序阶段一体化改进产品质量[J],控制世界钢铁,2005(6):33-42
[9]张晓丽,刘铁斐,刘淑杰等.基于C/S模式的数据仓库应用系统的研究[J],控制冶金自化,2007(4):67-73
[10]贾骥,数据仓库技术在宝钢1580mm热轧带钢生产中的应用尝试[J],控制宝钢技术,2006(5):33-36
[11]郑朝霞,刘廷建.关联规则在股票分析中的应用[J],控制成都大学学报(自然科学版),2007,21(4):46-49
[12]刘萍,别荣芳,基于交易数据库的关联规则生成算法FAS及其应用,计算机应用,2008,25(6):1376
[13]朱树人,李伟琴.ERP体系结构研究[J],控制系统工程,2004(3):40-43
[14]王媛媛,胡学钢.关联规则挖掘研究[J],控制计算机技术与应用进展,2008:808-813
[15]毕建欣,张岐山.关联规则挖掘算法综述[J],控制中国工程科学,2005,7(4):88-94
[16]曲守宁,董彩云,徐德军等.关联规则算法研究及其在教学系统中的应用[J],控制计算机系统应用,2007(4):20-23
[17]铁治欣,陈奇,俞瑞钊.关联规则采掘综述[J],控制计算机应用研究,2005,1( 1): 1-5
[18]罗可,吴杰.关联规则衡量标准的研究[J],控制控制与决策,2003, 18(3):277-280
[19]赛英.粗糙集扩展模型及其在数据挖掘中的应用研究:[博士学位论文].合肥:国防科学技术大学,2002
[20]李天瑞.数据库中的关联规则及挖掘算法研究:[博士学位论文].成都:西南交通大学,2006
[21] Verna Allee, The art and practice of being a revolutionary, journal ofnowledgeManagement, 1999, 3(2), 121-131
[22] P.Riches, J.Kemp and P.Wolf, Future of KM: Business Roadmap (Knowledge Management Transformation), European KM Forum,ST Project No:2000-26393
[23] KMQuick: AKM Tool for government practitioners, The FAA Knowledge Services Network and the Federal KM Network, 2002(8), Washington DC
[24] RSuresh, Knowledge Management: A Strategic Perspective, Source: The Provider、Edge
[25] H.Suresh, Knowledge Management:The road ahead for success, PSG Institute of Management, 2005(9)
[26] Knowledge management working group of the federal chief information officers Council, managing knowledge Mwork: An overview of knowledge management, CIO Council, 2007 (8)
[27] Karl-Erik Sveiby, Knowledge ManagementLessons from the Pioneer, NovemberDaniele Chauvel, Knowledge Management Models: A state of art
[28] Y.Malhotra, Knowledge mangement for e-business performance: Advancing information strategy to "Internet Time", Information Strategy, The Executive's journal, 2005, 16(4), 5-16
[29] Shu-Hsien Liao,Knowledge Management technologies and applications- literatureReview from 1995 to 2002, Expert System with Application, 2403(25), 155-164
[30] Polanyi.M, The tacit dimension, London: Rou.tledge and Kegan Paul
[31] Nonaka.I, Umemoto.K, & Senoo.D, from information processing to knowledge creating: a paradigm shift in business management, Technology in Society, 18(2), 203-218
[32] Leif Edvinsson, Developing a model for managing intellectual capital, European112 Management Journals, 2005(4), 356-364
[33] Jeef Wilkins, Understanding and valuing knowledge Assets: Overview and Method,Expert Systems with Applications,1997(13),55-72
[34] Karl M.Wiig, Supporting knowledge management: A selection of methods andtechniques,Expert Systems with Applications,1997(13),15-27
[35] Stephen Drew, Building knowledge management into strategy: Making sense of aNew perspective,Long Range Planning,1999(1),130-136
[36] B.Rubenstein-Montano, J.Liebowitz, J.Buchwalter, A system thinking frameworkFor knowledge management, Decision Support Systems, 2001(31), 5-16
[37] Jasbir S.Dhaliwal, Lai Lai Tung, Using group support systems for developinga knowledge-based explanation facility, International Journal of InformationManagement, 2000(20), 131-149
[38] Hsinchun Chen, Jenny Schroeder, Roslin Hauck, COPLINK Connect: information andKnowledge management for law enforcement, Decision Support Systems, 2002(34),271-285
[39] S.Cauvin, Dynamic application of action plans in the Alexip knowledge-basedSystem, Contrat Eng. Practice, 1996(1), 99-104
[40] KamelR ouibah,K evinR .Ca skey,Changem anagementi nconcurrente ngineeringf roma parameter perspective, Computers in Industry, 2003(50), 15-34
[41] C.Werner Dankwort, Roland Weidilich, Engineers' CAx education-it's not onlyCAD, Computers-Aided Design, 2004(36), 1439-1450
[42] Y. Malhotra, Why Knowledge management system fail? Enable and Constraint ofknowledgem anagementi nhumane nterprise,I nK .Srikantaiah&M.E. D. K oenig( Eds.),Knowledge management lessons learned: what works and what doesn't, Information Today(ASISTS eries),M edford,N .J. ,November.2005
[43] G. Fischer, Communities of interest: learning through the interaction ofmultiple knowledge systems, Proceedings of the 24"' IRIS conference, 2001, 1-14
[44]蔡林沁,谢阅,SPC数据采集与处理系统,计算机测量与控制,2002(5):675-677
[45]曾铁军,宋晓H,裴仁清,基于协同式专家系统的SPC诊断系统研制,中国机械工程,2002 (11):1852-1856
[46]王青,李明树.基于SPC的软件需求度量方法,计算机学报,2003(10):1312-1317
[47]陈卫斌,李虎山,王海儒等.连铸坯的热送热装工艺研究[J].太原重型机械学院学报,1998(4):321-325
[48]田乃媛.薄板坯连铸及热送直接轧制[M].北京:冶金工业出版社,1993
[49]殷瑞任.关于连铸与轧钢的衔接和匹配问题[J].轧钢,1991(2):1-7
[50]詹立新,杨自厚,王梦光.炼钢-连铸最优炉次计划模型与算法[J].东北大学学报(自然科学版),2003,17(4):440-445
[51]詹立新,杨自厚,王梦光.炼钢-连铸最优浇次(CAST)计划[J].东北大学学报(自然科学版),2003,17(5):554-558
[52]曾庆红.基于遗传算法求解炼钢-连铸浇次排序问题[J].鞍山师范学院学报,2004,6(6):70-73
[53]王莉.CIMS下的热轧生产批量计划的数学模型及解法[J].师范学院学报,1999,1(3):17-21
[54]严洪森,刘飞.知识化制造系统-新一代先进制造系统[J].计算机集成制造系统-CIMS,2001,7(8):7-11
[55]李靖.武钢连铸连轧生产物流一体化管理的研究[D].北京科技大学博士学位论文,2002
[56]殷瑞饪.流程工程学[M]。北京:冶金工业出版社,2004
[57]刘青,查晓燕,王良周等.炼钢厂的计算机辅助生产调度系统[J].钢铁,2O04,39(6):28-31
[58]刘青.现代长材型转炉炼钢厂的模式优化研究[D].北京科技大学博士学位论文,2002
[59]吕少华,边字编著,数据库开发与应用[M].北京:科学出版社,2001.6
[2]王立新.孢粉数据的回归方法[J].数理统计与管理,1995,14(4):2
[3]洪新杰.青霉素质量管理的统计方法[J].数理统计与管理,2005,14(4):2
[4]奚介荣,龚光荣,知识型制造企业与知识管理[J].机电一体化,2006(2):8-11
[5]王学仁,王松桂编译.实用统计分析[M].上海:上海科学技术出版杜,1999
[6]王学仁等.实用多元统计分析[M].上海:上海科学技术出版社,2000
[7]秘晓,鲍玉斌,常钊等.工业数据仓库设计方法及其在质量分析中的应用[J].控制与决策,2007(2):27-34
[8] Haald Peters(德)等.采用基于数据方法各工序阶段一体化改进产品质量[J],控制世界钢铁,2005(6):33-42
[9]张晓丽,刘铁斐,刘淑杰等.基于C/S模式的数据仓库应用系统的研究[J],控制冶金自化,2007(4):67-73
[10]贾骥,数据仓库技术在宝钢1580mm热轧带钢生产中的应用尝试[J],控制宝钢技术,2006(5):33-36
[11]郑朝霞,刘廷建.关联规则在股票分析中的应用[J],控制成都大学学报(自然科学版),2007,21(4):46-49
[12]刘萍,别荣芳,基于交易数据库的关联规则生成算法FAS及其应用,计算机应用,2008,25(6):1376
[13]朱树人,李伟琴.ERP体系结构研究[J],控制系统工程,2004(3):40-43
[14]王媛媛,胡学钢.关联规则挖掘研究[J],控制计算机技术与应用进展,2008:808-813
[15]毕建欣,张岐山.关联规则挖掘算法综述[J],控制中国工程科学,2005,7(4):88-94
[16]曲守宁,董彩云,徐德军等.关联规则算法研究及其在教学系统中的应用[J],控制计算机系统应用,2007(4):20-23
[17]铁治欣,陈奇,俞瑞钊.关联规则采掘综述[J],控制计算机应用研究,2005,1( 1): 1-5
[18]罗可,吴杰.关联规则衡量标准的研究[J],控制控制与决策,2003, 18(3):277-280
[19]赛英.粗糙集扩展模型及其在数据挖掘中的应用研究:[博士学位论文].合肥:国防科学技术大学,2002
[20]李天瑞.数据库中的关联规则及挖掘算法研究:[博士学位论文].成都:西南交通大学,2006
[21] Verna Allee, The art and practice of being a revolutionary, journal ofnowledgeManagement, 1999, 3(2), 121-131
[22] P.Riches, J.Kemp and P.Wolf, Future of KM: Business Roadmap (Knowledge Management Transformation), European KM Forum,ST Project No:2000-26393
[23] KMQuick: AKM Tool for government practitioners, The FAA Knowledge Services Network and the Federal KM Network, 2002(8), Washington DC
[24] RSuresh, Knowledge Management: A Strategic Perspective, Source: The Provider、Edge
[25] H.Suresh, Knowledge Management:The road ahead for success, PSG Institute of Management, 2005(9)
[26] Knowledge management working group of the federal chief information officers Council, managing knowledge Mwork: An overview of knowledge management, CIO Council, 2007 (8)
[27] Karl-Erik Sveiby, Knowledge ManagementLessons from the Pioneer, NovemberDaniele Chauvel, Knowledge Management Models: A state of art
[28] Y.Malhotra, Knowledge mangement for e-business performance: Advancing information strategy to "Internet Time", Information Strategy, The Executive's journal, 2005, 16(4), 5-16
[29] Shu-Hsien Liao,Knowledge Management technologies and applications- literatureReview from 1995 to 2002, Expert System with Application, 2403(25), 155-164
[30] Polanyi.M, The tacit dimension, London: Rou.tledge and Kegan Paul
[31] Nonaka.I, Umemoto.K, & Senoo.D, from information processing to knowledge creating: a paradigm shift in business management, Technology in Society, 18(2), 203-218
[32] Leif Edvinsson, Developing a model for managing intellectual capital, European112 Management Journals, 2005(4), 356-364
[33] Jeef Wilkins, Understanding and valuing knowledge Assets: Overview and Method,Expert Systems with Applications,1997(13),55-72
[34] Karl M.Wiig, Supporting knowledge management: A selection of methods andtechniques,Expert Systems with Applications,1997(13),15-27
[35] Stephen Drew, Building knowledge management into strategy: Making sense of aNew perspective,Long Range Planning,1999(1),130-136
[36] B.Rubenstein-Montano, J.Liebowitz, J.Buchwalter, A system thinking frameworkFor knowledge management, Decision Support Systems, 2001(31), 5-16
[37] Jasbir S.Dhaliwal, Lai Lai Tung, Using group support systems for developinga knowledge-based explanation facility, International Journal of InformationManagement, 2000(20), 131-149
[38] Hsinchun Chen, Jenny Schroeder, Roslin Hauck, COPLINK Connect: information andKnowledge management for law enforcement, Decision Support Systems, 2002(34),271-285
[39] S.Cauvin, Dynamic application of action plans in the Alexip knowledge-basedSystem, Contrat Eng. Practice, 1996(1), 99-104
[40] KamelR ouibah,K evinR .Ca skey,Changem anagementi nconcurrente ngineeringf roma parameter perspective, Computers in Industry, 2003(50), 15-34
[41] C.Werner Dankwort, Roland Weidilich, Engineers' CAx education-it's not onlyCAD, Computers-Aided Design, 2004(36), 1439-1450
[42] Y. Malhotra, Why Knowledge management system fail? Enable and Constraint ofknowledgem anagementi nhumane nterprise,I nK .Srikantaiah&M.E. D. K oenig( Eds.),Knowledge management lessons learned: what works and what doesn't, Information Today(ASISTS eries),M edford,N .J. ,November.2005
[43] G. Fischer, Communities of interest: learning through the interaction ofmultiple knowledge systems, Proceedings of the 24"' IRIS conference, 2001, 1-14
[44]蔡林沁,谢阅,SPC数据采集与处理系统,计算机测量与控制,2002(5):675-677
[45]曾铁军,宋晓H,裴仁清,基于协同式专家系统的SPC诊断系统研制,中国机械工程,2002 (11):1852-1856
[46]王青,李明树.基于SPC的软件需求度量方法,计算机学报,2003(10):1312-1317
[47]陈卫斌,李虎山,王海儒等.连铸坯的热送热装工艺研究[J].太原重型机械学院学报,1998(4):321-325
[48]田乃媛.薄板坯连铸及热送直接轧制[M].北京:冶金工业出版社,1993
[49]殷瑞任.关于连铸与轧钢的衔接和匹配问题[J].轧钢,1991(2):1-7
[50]詹立新,杨自厚,王梦光.炼钢-连铸最优炉次计划模型与算法[J].东北大学学报(自然科学版),2003,17(4):440-445
[51]詹立新,杨自厚,王梦光.炼钢-连铸最优浇次(CAST)计划[J].东北大学学报(自然科学版),2003,17(5):554-558
[52]曾庆红.基于遗传算法求解炼钢-连铸浇次排序问题[J].鞍山师范学院学报,2004,6(6):70-73
[53]王莉.CIMS下的热轧生产批量计划的数学模型及解法[J].师范学院学报,1999,1(3):17-21
[54]严洪森,刘飞.知识化制造系统-新一代先进制造系统[J].计算机集成制造系统-CIMS,2001,7(8):7-11
[55]李靖.武钢连铸连轧生产物流一体化管理的研究[D].北京科技大学博士学位论文,2002
[56]殷瑞饪.流程工程学[M]。北京:冶金工业出版社,2004
[57]刘青,查晓燕,王良周等.炼钢厂的计算机辅助生产调度系统[J].钢铁,2O04,39(6):28-31
[58]刘青.现代长材型转炉炼钢厂的模式优化研究[D].北京科技大学博士学位论文,2002
[59]吕少华,边字编著,数据库开发与应用[M].北京:科学出版社,2001.6