工业系统集成信息处理技术及其应用研究
|
摘要
本文的目的是为了解决工业系统中现存的数据采集与数据利用之间的矛盾，从工业系统大量的原始数据中发现隐藏的知识，建立信息处理模型，从而采取相应的智能行为，组成性能更高的系统，为那些生产过程机理复杂、不易建立精确数学模型的工业系统提供一种基于其自身数据的建模方式。
论文以系统的观点作为指导，探索出的“数据—信息—知识—智能行为”转化关系，揭示了工业系统信息处理的基本转化过程，在这一框架之下开展了后续的研究工作；详细分析了数据挖掘技术与数据融合技术的异同与互补之处，指出了这两种技术在工业系统信息处理过程中的地位，以及单独应用的局限性，提出了数据挖掘与数据融合相集成的信息处理技术，给出了具体实现模型；独立于硬件结构，从信息处理角度抽象出了工业系统的拓扑结构；对信息处理系统的输入输出进行了分类，提出了相应的处理方法；提出了基于集成信息处理技术的工业系统信息处理总体模型，分析了其中抽象层次与信息量解析水准的关系，探讨了软硬件实现的基本方法.
论文将模型应用于冶金企业焦化厂炼焦生产过程，利用人工神经网络集成、粗糙集理论等计算智能技术作为具体实现算法，在原有监测监控系统的基础上，编制了软件程序，实现了焦炭质量预测和焦炉炉温的智能控制，为这两个长期困扰炼焦生产的难题提供了一种新的解决方法；在具体应用的基础上，总结出了建立工业信息处理总体模型的方法和步骤。
The purpose of this thesis is to resolve the puzzle of the contradiction between data acquisition and data utilization in the industry system, to develop methods in finding hidden knowledge from vast primary data and developing information process models accordingly. Thereby the industry system’s performance could be improved by taking corresponding intelligence actions. And a modeling method based on the data origin from the industry system itself could be adopted if the precise math model is hard to develop because of the complicated production process.
Research works are taken under the guidance of system analysis. First, the conversion relations of “data—information—knowledge—intelligence actions” are detected out, which uncover the basic tasks of the industry system information process. Further research works run under these relations of conversion. Second, as to data mining and data fusion, particular analyses are done in seeking the similarity and difference between these two. Some reciprocity of these two is found. Functions and limitations of respectively using of these two in industry system information process are also pointed out. With this understanding, an integrated information process technique and the corresponding models are developed. Third, the industry system’s topological structure, which independent from hardware facilities, is abstracted at the angle of information process. The input and the output of the information process system are classified, and corresponding process methods are developed. Then the general information process model of industry system is developed based on the integrated information process technique. After that, the relations between the model’s abstract levels and analytic levels of the amount of information are discussed, and the basic implementation methods are discussed too. Furthermore, these models are applied in the coking process of the coking factory in a metallurgical enterprise. In this case, artificial neural network ensembles and rough set theory are applied as the concrete algorithm; software programs are designed on the condition of the existent of monitor and supervise control system. As a result of the using
of the industry system’s general information process model, the factory’s long-puzzled problem of coke quality forecasting and coke oven temperature’s intelligent control are basically resolved. Finally, the steps of developing a general model of the industry information process are concluded.
论文以系统的观点作为指导，探索出的“数据—信息—知识—智能行为”转化关系，揭示了工业系统信息处理的基本转化过程，在这一框架之下开展了后续的研究工作；详细分析了数据挖掘技术与数据融合技术的异同与互补之处，指出了这两种技术在工业系统信息处理过程中的地位，以及单独应用的局限性，提出了数据挖掘与数据融合相集成的信息处理技术，给出了具体实现模型；独立于硬件结构，从信息处理角度抽象出了工业系统的拓扑结构；对信息处理系统的输入输出进行了分类，提出了相应的处理方法；提出了基于集成信息处理技术的工业系统信息处理总体模型，分析了其中抽象层次与信息量解析水准的关系，探讨了软硬件实现的基本方法.
论文将模型应用于冶金企业焦化厂炼焦生产过程，利用人工神经网络集成、粗糙集理论等计算智能技术作为具体实现算法，在原有监测监控系统的基础上，编制了软件程序，实现了焦炭质量预测和焦炉炉温的智能控制，为这两个长期困扰炼焦生产的难题提供了一种新的解决方法；在具体应用的基础上，总结出了建立工业信息处理总体模型的方法和步骤。
The purpose of this thesis is to resolve the puzzle of the contradiction between data acquisition and data utilization in the industry system, to develop methods in finding hidden knowledge from vast primary data and developing information process models accordingly. Thereby the industry system’s performance could be improved by taking corresponding intelligence actions. And a modeling method based on the data origin from the industry system itself could be adopted if the precise math model is hard to develop because of the complicated production process.
Research works are taken under the guidance of system analysis. First, the conversion relations of “data—information—knowledge—intelligence actions” are detected out, which uncover the basic tasks of the industry system information process. Further research works run under these relations of conversion. Second, as to data mining and data fusion, particular analyses are done in seeking the similarity and difference between these two. Some reciprocity of these two is found. Functions and limitations of respectively using of these two in industry system information process are also pointed out. With this understanding, an integrated information process technique and the corresponding models are developed. Third, the industry system’s topological structure, which independent from hardware facilities, is abstracted at the angle of information process. The input and the output of the information process system are classified, and corresponding process methods are developed. Then the general information process model of industry system is developed based on the integrated information process technique. After that, the relations between the model’s abstract levels and analytic levels of the amount of information are discussed, and the basic implementation methods are discussed too. Furthermore, these models are applied in the coking process of the coking factory in a metallurgical enterprise. In this case, artificial neural network ensembles and rough set theory are applied as the concrete algorithm; software programs are designed on the condition of the existent of monitor and supervise control system. As a result of the using
of the industry system’s general information process model, the factory’s long-puzzled problem of coke quality forecasting and coke oven temperature’s intelligent control are basically resolved. Finally, the steps of developing a general model of the industry information process are concluded.
引文
[1]Fayyad U, Piatetsky-Shapiro, Smyth, Uthurusamy. Advances in Knowledge Discovery and Data Mining. MIT Press. 1996
[2]Usama Fayyad, Paul Stolorz. Data mining and KDD: Promise and challenges. Future。Generation Computer Systems. November, 1997, Vol: 13, 99-115
[3]黄绍君等.知识发现及其应用研究回顾.计算机应用研究,2001,4:1~8.
[4]马元元等.增量关联规则在大型火力发电厂实时控制中的应用.工业控制计算机2000,13(1):14~15
[5]杨会志等.基于RS理论的过程企业生产工艺数据挖掘.计算机工程,2001,27(11):37~38
[6]卢勇等数据挖掘技术在热电厂过程控制与优化中的应用研究.电站系统工程,2003,19(2):49~50
[7]俞章毅等. 数据挖掘技术及其在化工过程中的应用.浙江化工,2003,9:26~27.
[8]杨学兵等. 用分类树的方法探索最佳工艺.华东冶金学院学报,2000,17(2):159~163
[9]张邦礼等.基于粗糙集理论的内燃机气阀故障诊断研究.内燃机学报2002,20(2):153~156
[10]杨杰等.数据挖掘技术在建模、优化和故障诊断中的应用.红外与激光工程2000,29(3):23~27
[11]石金彦等.粗糙集与决策树结合诊断故障的数据挖掘方法.郑州大学学报(工学版)，2003,24(1):109~112
[12]刘丽萍等.自动化加工过程的在线质量监测及管理平台的研究 工业工程与管理,2003,1:42~46
[13]刘士新等.钢铁企业产成品预报模型.东北大学学报(自然科学版)2003,24(3):241~243
[14]梅时春等.过程监控中数据挖掘与知识发现理论及应用.微计算机信息2002,18(2):1~3
[15]罗印升等.复杂工业过程中数据挖掘模型研究.信息与控制,2003,32(1):32~35
[16]杨学兵等.一种实时控制中的数据挖掘算法研究.计算机应用,1999,19(9):8~10
[17]刘同明等.数据融合技术及应用.北京:国防工业出版社，2000.1~2
[18]Special Issue on data fusion. Proceedings of The IEEE.1997,85(1)
[19]葛非等.基于人工神经网络(ANN)的智能检测系统数据融合研究.科技进步与对策2002,17(12):179~180
[20]罗志增等.机器人感觉与多信息融合.北京:机械工业出版社,2002.8~9,95~110
[21]孟宪尧等.贝叶斯数据融合技术在机舱故障智能诊断中的应用.大连海事大学学报2002,28(3):10~13
[22]刘燕燕,等.数据融合技术在输电线网故障诊断中的应用. 信息技术,2002,8:2~5
[23]姜万录,等.多传感器数据融合技术的现状及展望.机床与液压,2003,3:16~18
[24]史忠植．知识发现.北京:清华大学出版社,2002.1~15
[25]顾培亮.系统分析与协调.天津:天津大学出版社,1998.22~34,35~36,41~45
[26]Vapink,V.Statistical Learning Theory. Addison-Wisley,1998
[27]Pawlak Z.Rough sets.International journal of Computer and information Sciences. 1982(11):341～356
[28]Durrant-Whyte. Sensor Models and Multisensor Intergration. Jour of Robotics Research, 1988,7(6): 97~100
[29]权太范.信息融合:神经网络—模糊推理理论与应用.北京:国防工业出版社,2002.48~49
[30]Fogel David B.Toshio Fukuda.Ling Guan.Special Issue on Computational Intelligence.Proceedings of the IEEE,1999,87( 9):1415~1422
[31]W.McCulloch and W.Pitts, “A logical calculus of the ideas immanent in nervous activity,” Bulletin of Mathematical Biophysis., 1943,Vol.5:115~133
[32]易继锴,等.智能控制技术.北京:北京工业大学,1999.95~108
[33]ValiantLG.ATheoryoftheLearnable.Communications of the ACM, 1984,27(11):1134~1142
[34]Hansen L K, Salamon P. Neural Network Ensembles.IEEE Transactions on Pattern Analysis and Machine Intelligence, 1990, 12(10):993~1001
[35]Sollich P, Krogh A. Learning with Ensembles: How Over-Fitting Can Be Useful. In:Touretzky D,Mozer M, Hasselmo M(eds.) Advances in Neural Information Processing System (Volume 8),Cambridge,MA: MIT Press,1996.190~196
[36]陈兆乾等.神经计算研究现状及发展趋势.见:路汝钤编.知识科学与计算科学.北京:清华大学出版社,2003.165~177
[37]吴建鑫等.一种选择性神经网络集成构造方法.计算机研究与发展2000,37(9):1039~1044
[38]张文修等.粗糙集理论与方法.北京:科学出版社,2001.1~39
[39]王国胤.Rough理论与知识获取.西安:西安交通大学出版社,2001.14~52,92~152
[40]钟义信.“知识论”基础研究.电子学报,2001,29(1):98~102
[41]钟义信.知识论:核心问题—信息-知识-智能的统一理论.电子学报2001,29(4):526~530
[42]杨世兴等.监测监控系统.西安,西安出版社,1999.1~11
[43]叶道敏.煤岩配煤和焦炭强度的预测.燃料与化工,2003,29(6):233~236
[44]张纪民等.高炉煤气加热智能系统在焦炉的应用.燃料与化工,2003,34(1):20~22
[45]Waltz,Edward L,Information understanding:Integrating datafusion and data minging processes, proceedings-IEEE International Symposium on circuits and
systems,v6,1998,p553~556
[2]Usama Fayyad, Paul Stolorz. Data mining and KDD: Promise and challenges. Future。Generation Computer Systems. November, 1997, Vol: 13, 99-115
[3]黄绍君等.知识发现及其应用研究回顾.计算机应用研究,2001,4:1~8.
[4]马元元等.增量关联规则在大型火力发电厂实时控制中的应用.工业控制计算机2000,13(1):14~15
[5]杨会志等.基于RS理论的过程企业生产工艺数据挖掘.计算机工程,2001,27(11):37~38
[6]卢勇等数据挖掘技术在热电厂过程控制与优化中的应用研究.电站系统工程,2003,19(2):49~50
[7]俞章毅等. 数据挖掘技术及其在化工过程中的应用.浙江化工,2003,9:26~27.
[8]杨学兵等. 用分类树的方法探索最佳工艺.华东冶金学院学报,2000,17(2):159~163
[9]张邦礼等.基于粗糙集理论的内燃机气阀故障诊断研究.内燃机学报2002,20(2):153~156
[10]杨杰等.数据挖掘技术在建模、优化和故障诊断中的应用.红外与激光工程2000,29(3):23~27
[11]石金彦等.粗糙集与决策树结合诊断故障的数据挖掘方法.郑州大学学报(工学版)，2003,24(1):109~112
[12]刘丽萍等.自动化加工过程的在线质量监测及管理平台的研究 工业工程与管理,2003,1:42~46
[13]刘士新等.钢铁企业产成品预报模型.东北大学学报(自然科学版)2003,24(3):241~243
[14]梅时春等.过程监控中数据挖掘与知识发现理论及应用.微计算机信息2002,18(2):1~3
[15]罗印升等.复杂工业过程中数据挖掘模型研究.信息与控制,2003,32(1):32~35
[16]杨学兵等.一种实时控制中的数据挖掘算法研究.计算机应用,1999,19(9):8~10
[17]刘同明等.数据融合技术及应用.北京:国防工业出版社，2000.1~2
[18]Special Issue on data fusion. Proceedings of The IEEE.1997,85(1)
[19]葛非等.基于人工神经网络(ANN)的智能检测系统数据融合研究.科技进步与对策2002,17(12):179~180
[20]罗志增等.机器人感觉与多信息融合.北京:机械工业出版社,2002.8~9,95~110
[21]孟宪尧等.贝叶斯数据融合技术在机舱故障智能诊断中的应用.大连海事大学学报2002,28(3):10~13
[22]刘燕燕,等.数据融合技术在输电线网故障诊断中的应用. 信息技术,2002,8:2~5
[23]姜万录,等.多传感器数据融合技术的现状及展望.机床与液压,2003,3:16~18
[24]史忠植．知识发现.北京:清华大学出版社,2002.1~15
[25]顾培亮.系统分析与协调.天津:天津大学出版社,1998.22~34,35~36,41~45
[26]Vapink,V.Statistical Learning Theory. Addison-Wisley,1998
[27]Pawlak Z.Rough sets.International journal of Computer and information Sciences. 1982(11):341～356
[28]Durrant-Whyte. Sensor Models and Multisensor Intergration. Jour of Robotics Research, 1988,7(6): 97~100
[29]权太范.信息融合:神经网络—模糊推理理论与应用.北京:国防工业出版社,2002.48~49
[30]Fogel David B.Toshio Fukuda.Ling Guan.Special Issue on Computational Intelligence.Proceedings of the IEEE,1999,87( 9):1415~1422
[31]W.McCulloch and W.Pitts, “A logical calculus of the ideas immanent in nervous activity,” Bulletin of Mathematical Biophysis., 1943,Vol.5:115~133
[32]易继锴,等.智能控制技术.北京:北京工业大学,1999.95~108
[33]ValiantLG.ATheoryoftheLearnable.Communications of the ACM, 1984,27(11):1134~1142
[34]Hansen L K, Salamon P. Neural Network Ensembles.IEEE Transactions on Pattern Analysis and Machine Intelligence, 1990, 12(10):993~1001
[35]Sollich P, Krogh A. Learning with Ensembles: How Over-Fitting Can Be Useful. In:Touretzky D,Mozer M, Hasselmo M(eds.) Advances in Neural Information Processing System (Volume 8),Cambridge,MA: MIT Press,1996.190~196
[36]陈兆乾等.神经计算研究现状及发展趋势.见:路汝钤编.知识科学与计算科学.北京:清华大学出版社,2003.165~177
[37]吴建鑫等.一种选择性神经网络集成构造方法.计算机研究与发展2000,37(9):1039~1044
[38]张文修等.粗糙集理论与方法.北京:科学出版社,2001.1~39
[39]王国胤.Rough理论与知识获取.西安:西安交通大学出版社,2001.14~52,92~152
[40]钟义信.“知识论”基础研究.电子学报,2001,29(1):98~102
[41]钟义信.知识论:核心问题—信息-知识-智能的统一理论.电子学报2001,29(4):526~530
[42]杨世兴等.监测监控系统.西安,西安出版社,1999.1~11
[43]叶道敏.煤岩配煤和焦炭强度的预测.燃料与化工,2003,29(6):233~236
[44]张纪民等.高炉煤气加热智能系统在焦炉的应用.燃料与化工,2003,34(1):20~22
[45]Waltz,Edward L,Information understanding:Integrating datafusion and data minging processes, proceedings-IEEE International Symposium on circuits and
systems,v6,1998,p553~556