基于规则的数据挖掘方法在故障诊断中的应用
|
摘要
针对数据库和数据仓库技术在故障诊断领域的广泛应用以及现在各工厂基本上都对重要设备实施了实时监控，由传感器不间断的传回反映机组运行状态的各种数据及参数，已经形成大型的数据库或数据仓库这一事实，提出将数据挖掘中产生显式规则的方法应用于该领域。决策树、粗糙集、关联规则等数据挖掘方法能产生显示的规则，并能有效解决海量数据中知识的发现问题。
决策树是一个类似于流程图的树结构，主要用途是提取分类规则，进行分类预测。粗糙集理论无需提供除问题相关的数据集合外的任何先验信息，适合于发现数据中潜在的规律、不准确数据或噪声数据内在的结构联系，可以解决重要的分类问题。关联规则反映一个事件和其他事件之间依赖或关联的知识。如果两项或多项属性之间存在关联，那么其中一项的属性值就可以依据其他属性值进行预测。应用于故障诊断领域，对大量的机组状态数据进行挖掘，发现故障数据中存在的规律，以规则的形式体现出来，可以为故障诊断提供决策依据。利用转子实验台的模拟故障数据对提出的方法进行考核。结果表明，系统采用的几种方法所得出的规则是正确的，并能正确用于故障的分类工作，也可为专家提供有价值的信息。
在WINDOWS98开发平台上，结合开发工具Microsoft Visual C++ 6.0，SQL Server数据库管理系统。采用面向对象的程序设计思想和模块化程序设计方法，对基于显式规则的几种数据挖掘方法进行了软件实现，开发了基于故障诊断的挖掘系统，该系统具有一定的实用价值。
本文的创新点在于将数据挖掘产生规则的方法引入了故障诊断，结合先进的数据库管理系统，将数据挖掘中关联规则、粗糙集以及决策树方法用于故障的分类工作，并最终软件实现了以上的构想。
In accordance with the widely application of databases and data warehouses, and the reality of the installation of on-line and off-line monitoring system to significant equipment and large-scale databases and data warehouses have come into being hi fault diagnosis field, the paper represents a new method of data mining which produces visible rules. Potential knowledge can be effectively discovered by data mining method such as decision tree, rough set and association rules from a mass of fault data.
Decision tree, as a flow chart, is structure of a tree, which is mostly used in finding classification rules and prediction of classification. Rough set theory is mainly used in attributes reduction and classification. Potential rule, internal relation of inaccurate or noise data can be discovered by rough set, although there is no any prior knowledge expect the data set associated with actual problem. Association rules reflect the knowledge relied or associated between one event and the other events. If associations between two attributes or several attributes are got hold, one attribute value among these attributes can be predicted by values of other attributes. When such data mining methods are used hi fault diagnosis field based on a mass of fault data about machinery state, significant information can be discovered and be showed as visible rules and decisive conclusions for diagnosis be acquired from these rules. The results indicate that rules generated by data mining system are according with actual fault features and this system can be used hi accurate classification of different faults.
hi the environment of Microsoft Visual C++6.0 and Database Management System (DBMS) SQL ServerT.O, the Chinese version of Window98 operating system, a fault diagnosis system is developed by using data mining methods, hi which visible rules are produced Finally, the system is tested by experiment data of a rotaiy machinery.
The innovative points of this paper are as followed: introducing the data mining methods which produce visible rules to fault diagnosis, developing the fault diagnosis system based on data mining combining with advanced database management system, using association rules and rough set and decision tree to classify faults.
决策树是一个类似于流程图的树结构，主要用途是提取分类规则，进行分类预测。粗糙集理论无需提供除问题相关的数据集合外的任何先验信息，适合于发现数据中潜在的规律、不准确数据或噪声数据内在的结构联系，可以解决重要的分类问题。关联规则反映一个事件和其他事件之间依赖或关联的知识。如果两项或多项属性之间存在关联，那么其中一项的属性值就可以依据其他属性值进行预测。应用于故障诊断领域，对大量的机组状态数据进行挖掘，发现故障数据中存在的规律，以规则的形式体现出来，可以为故障诊断提供决策依据。利用转子实验台的模拟故障数据对提出的方法进行考核。结果表明，系统采用的几种方法所得出的规则是正确的，并能正确用于故障的分类工作，也可为专家提供有价值的信息。
在WINDOWS98开发平台上，结合开发工具Microsoft Visual C++ 6.0，SQL Server数据库管理系统。采用面向对象的程序设计思想和模块化程序设计方法，对基于显式规则的几种数据挖掘方法进行了软件实现，开发了基于故障诊断的挖掘系统，该系统具有一定的实用价值。
本文的创新点在于将数据挖掘产生规则的方法引入了故障诊断，结合先进的数据库管理系统，将数据挖掘中关联规则、粗糙集以及决策树方法用于故障的分类工作，并最终软件实现了以上的构想。
In accordance with the widely application of databases and data warehouses, and the reality of the installation of on-line and off-line monitoring system to significant equipment and large-scale databases and data warehouses have come into being hi fault diagnosis field, the paper represents a new method of data mining which produces visible rules. Potential knowledge can be effectively discovered by data mining method such as decision tree, rough set and association rules from a mass of fault data.
Decision tree, as a flow chart, is structure of a tree, which is mostly used in finding classification rules and prediction of classification. Rough set theory is mainly used in attributes reduction and classification. Potential rule, internal relation of inaccurate or noise data can be discovered by rough set, although there is no any prior knowledge expect the data set associated with actual problem. Association rules reflect the knowledge relied or associated between one event and the other events. If associations between two attributes or several attributes are got hold, one attribute value among these attributes can be predicted by values of other attributes. When such data mining methods are used hi fault diagnosis field based on a mass of fault data about machinery state, significant information can be discovered and be showed as visible rules and decisive conclusions for diagnosis be acquired from these rules. The results indicate that rules generated by data mining system are according with actual fault features and this system can be used hi accurate classification of different faults.
hi the environment of Microsoft Visual C++6.0 and Database Management System (DBMS) SQL ServerT.O, the Chinese version of Window98 operating system, a fault diagnosis system is developed by using data mining methods, hi which visible rules are produced Finally, the system is tested by experiment data of a rotaiy machinery.
The innovative points of this paper are as followed: introducing the data mining methods which produce visible rules to fault diagnosis, developing the fault diagnosis system based on data mining combining with advanced database management system, using association rules and rough set and decision tree to classify faults.
引文
[1] 韩捷、张瑞林等．旋转机械故障机理及诊断技术．北京：机械工业出版社．1997．8．5-19
[2] 孙惠国、张瑞林．设备状态检测与故障诊断．郑州：河南科学技术出版社 1993
[3] 陈大禧、朱铁光．大型回转机械诊断现场实用技术．北京：机械工业出版社．3-90
[4] 钟秉林、黄仁．机械故障诊断学．北京：机械工业出版社．1997．12．29-52
[5] 唐常杰、杨富华、杨璐．数据挖掘的基本方法及其与专家系统的差异．计算机应用．1999．3．17-21
[6] 姜东．基于数据挖掘的故障诊断系统[硕士学位论文]．2001．5．1-10
[7] 黄德双．神经网络模式识别系统理论．北京：电子工业出版社．1996．5．12-117
[8] 崔锦泰[美]、程正兴．小波分析导论．西安：西安交通大学出版社．1995．1．2-24
[9] 胡侃、夏绍伟．基于大型数据仓库的数据采掘：研究综述．软件学报．1998．1．53-63
[10] 陈宁、周龙骧．数据采掘技术．计算机世界(技术专题)第19期D版．1998．5
[11] 郭萌、王珏，数据挖掘与数据库知识发现：综述，模式识别与人工智能，1998．3
[12] 朱廷劭、王军．KDD：数据挖掘应用．计算机世界(技术专题)第37期D版．1998．9
[13] 田金兰、李奔．数据挖掘工具：分类器．计算机世界(产品与技术)第20期C版．1999．5
[14] 冯铎．数据挖掘技术概述．电子周刊．http://www.China-Pub.com 2001.4.
[15] 沈清、汤霖．模式识别导论．长沙：国防科技大学出版社．1991．5．54-101
[16] 屈梁生、何正嘉．机械故障诊断学．上海：上海科技出版社．1986．3
[17] 石金彦、黄士涛、雷文平．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．2003．3．109-112
[18] Rasoul S, David Landgrebe. A Survey of Decision Tree Classifier Methodology. IEEE Transactions on Systems, Man and Cybernetics. 1991.5. 660-674
[19] Jiawei Hart、Micheline Hamber著，范明、孟小峰译．数据挖掘概念与技术．北京：机械工业出版社．2001．8．70-195
[20] Ouinlan J.R. Simplifying Decision Trees. Intemational Journal of Man-Machine Studies. 1987.3.221-248.
[21] Quinlan J.R. Induction of Decision Trees. Machine Learning. 1986.1.81-106
[22] Quinlan J.R. Improved Use of Continuous Attributes in C4.5. Journal of Artificial Intelligence Research. 1996.3.77-90.
[23] Dash M, Liu H. Feature Selection for Classification. Intelligent Data Analysis. 1997.1. 131-156
[24] Rissanen J、Wax M. Algorithm for Constructing Tree Structured Classifiers. US Patent. No.4719571.1988
[25] Mingers J. An Empirical Comparison of Pruning Methods for Decision Tree Induction,
Machine Learning. 1989.4. 227-243
[26] Oates T, Jensen D. The E fleets of Training Set Size On Decision Tree Complexity. Proceeding of the 14th International Conference On Machine Learning. 1997. 265-262
[27] Terri Moore, Carole Jesse、Riehard Kittler. An Overview and Evaluation of Decision Tree Methodology. ASA Quality and Productivity Conference. 2001.5.23-25
[28] Hess, Sidney W. Swinging on Branch of a Tree: Project Selection Application. Interfaces. 1993.12. 5-12
[29] Quinlan JR. C4.5: Programming for Machine Learning. CA. 1993
[30] Gunther Gediga, Ivo Duntseh. Statistical Tools for Rule Based Data Analysis. Synthesis of Intelligent Agent Systems From Experimental. ECAI98. 1998. 173-242
[31] Agrawal, Imielinski, Swami. Database Mining: A Performance Perspective. IEEE Transactions On Knowledge and Data Engineering. 1993.5. 914-925
[32] 刘同明．数据挖掘技术及其应用．北京：国防工业出版社，2001．9
[33] DanL Grecu, LeeA Becker. Coactive Learning for Distributed Data Mining. New York: 4th International Conference on Knowledge Discovery and Data Mining. 1998.7.
[34] Bhandari, Colet, Parker. Advanced Scout: Data Mining and Knowledge Discovery in NBA Data. Data Mining and Knowledge Discovery. 1997.1.121-125
[35] David Heckerman. Bayesian Networks for Data Mining. Data Mining and Knowledge Discovery. 1997.1.79-119
[36] 王国胤．Rough集理论与知识获取．西安：西安交通大学出版社．2001．5
[37] Z. Pawlak. Rough Sets. Int'I J. Compute & Science. 1982.10. 341-356
[38] Z. Pawlak. Rough Set Theory and Its Applications To Data Analysis. Cybernetics & Systems: An Int'I J. 1998. 661-688
[39] Ivo Dutseh, Gunter Gediga. The Rough Set Model for Data Analysis: Introduction and Overview. Preprint. http://www.infj.ulst.aciuk/～cccz23/papers/roughmod.html.1996.7
[40] Ivo Dutsch, Gunter Gediga. The Rough Set Engine GROBIAN. Seheol of Information and software Engineering. http://www.irfj.ulst.aciuk/～cccz23/papers/grobian.pdf
[41] Ziarko W, Shan. Rough Sets as a Methodology in Data Mining. 1998. 554-573
[42] 王钰、苗夺谦、周育健．关于Rough Set理论及应用的综述．模式识别与人工智能． 1996．12．337-344
[43] Ziarko W. Variable Precision Rough Set Model. Journal of Computer and System Sciences. 1993.46
[44] Ziarko W. AM ethod for Computing all Maximally General Rules In Attribute-value S.ystems. Computational Intelligence. 1993.2.2-13
[45] Bell D, Guan J. Computational Methods for Rough Classification and Discovery. Journal of ASIS 49. 1998.5.403-414
[46] Deogun J, Raghavan V, Sever H. Rough Set based classification methods and extended decision tables. In Proceedings of the International Workshop On Rough Sets and Soft
Computing. 1994. 302-309
[47] Sever H、Raghavan V. The Status of Research On Rough Sets For Knowledge Discovery In Databases. In Proceedings of the International Workshop On Rough Sets and Soft Computing. 1994. 325-332
[48] 常犁云、王国胤、吴渝．一种基于R0ugll Set理论的属性约简及规则提取方法．软件学报．1999．11．1206-1211
[49] 高毅龙．数据挖掘及其在工程诊断中的应用[博士学位论文]．2000．3
[50] Agrawal R, Imielienski T、Swami A. Mining Association Rules Between Sets of Items In Large Databases. Proc. Conf. On Management of Data. New York: ACM press. 1993. 10. 207-216
[51] Agrawal R, Mannila H、Srikant R. Fast Discovery of Association Rules. In: Fayyad et al. 1996. 307-328
[52] Hannu Toivonen. Sampling Large Databases for Association Rules. Proceeding of the 22nd VLDB Conference Mumbai(Bombay). India. 1996
[53] Zheng、Kohavi R、Mason L. Real World Performance of. Association Rule Algorithms. In: Proc. 7th Int. Conf On Knowledge Discovery and Data Mining(SIGKDD'01). 2001
[54] Mazlack J. Granulation of Quantitative Association Rules. International Journal of Fuzzy Systems. 2001.6. 400-406
[55] Brin S、Motwani R. Beyond Market Basket: Generalizing Association Rules To Correlations. In Proceeding of the ACM SIGMOD International Conference on Management of Data(ACM SIGMOD'97). 1997a. 265-276
[56] Bing Liu、Wynne Hsu、Yiming Ma. Integrating Classification and Association Rule Mining. KDD98. 1998.7.27-31
[57] Fukada T、Morimoto Y、Morishita S. Mining Optimized Association Rules for Numeric Attributes. In Proceeding of 15th ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems(PODS'96). 1996. 182-191
[58] Bing Liu、Wymae Hsu、Yiming Ma. Mining Association Rules with Multiple Minimum Supports. ACM SIGKDD International Conference On Knowledge Discovery & Data Mining (KDD'99). 1999.7
[59] Hipp J、Myka A、Wirth R. A New Algorithm for Faster Mining of Generalized Association Rules. In Proceedings of the 2nd European Symposium on Principles of Data Mining and Knowledge Discovery(PKDD'98). 1998.74-82
[60] Brin S、Motwani R. Dynamic Item Set Counting and Implication Rules for Market Basket Data. In Proceeding of the ACM SIGMOD International Conference on Management of Data(ACM SIGMOD'97). 1997b. 265-276
[61] Bing Liu、Wynne Hsu、Yiming Ma. Pruning and Summarizing the Discovered Associations. KDD'99. 1999.7
[62] 杨文献、任兴民．基于数据挖掘的复杂设备故障诊断方法研究．振动工程学报．
2000．10
[63] 李峰．数据库技术在汽轮机故障诊断系统中的应用．动力工程．第20卷．第1期．2000．2．549-553
[64] David Bennett．徐军等译．Visual C++5.0开发人员指南．北京：机械工业出版社．
[65] Richard C、Leinceker．陈冠民、王如亮等译．Visual C++5.0开发技术内幕．北京：机械工业出版社．
[2] 孙惠国、张瑞林．设备状态检测与故障诊断．郑州：河南科学技术出版社 1993
[3] 陈大禧、朱铁光．大型回转机械诊断现场实用技术．北京：机械工业出版社．3-90
[4] 钟秉林、黄仁．机械故障诊断学．北京：机械工业出版社．1997．12．29-52
[5] 唐常杰、杨富华、杨璐．数据挖掘的基本方法及其与专家系统的差异．计算机应用．1999．3．17-21
[6] 姜东．基于数据挖掘的故障诊断系统[硕士学位论文]．2001．5．1-10
[7] 黄德双．神经网络模式识别系统理论．北京：电子工业出版社．1996．5．12-117
[8] 崔锦泰[美]、程正兴．小波分析导论．西安：西安交通大学出版社．1995．1．2-24
[9] 胡侃、夏绍伟．基于大型数据仓库的数据采掘：研究综述．软件学报．1998．1．53-63
[10] 陈宁、周龙骧．数据采掘技术．计算机世界(技术专题)第19期D版．1998．5
[11] 郭萌、王珏，数据挖掘与数据库知识发现：综述，模式识别与人工智能，1998．3
[12] 朱廷劭、王军．KDD：数据挖掘应用．计算机世界(技术专题)第37期D版．1998．9
[13] 田金兰、李奔．数据挖掘工具：分类器．计算机世界(产品与技术)第20期C版．1999．5
[14] 冯铎．数据挖掘技术概述．电子周刊．http://www.China-Pub.com 2001.4.
[15] 沈清、汤霖．模式识别导论．长沙：国防科技大学出版社．1991．5．54-101
[16] 屈梁生、何正嘉．机械故障诊断学．上海：上海科技出版社．1986．3
[17] 石金彦、黄士涛、雷文平．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．2003．3．109-112
[18] Rasoul S, David Landgrebe. A Survey of Decision Tree Classifier Methodology. IEEE Transactions on Systems, Man and Cybernetics. 1991.5. 660-674
[19] Jiawei Hart、Micheline Hamber著，范明、孟小峰译．数据挖掘概念与技术．北京：机械工业出版社．2001．8．70-195
[20] Ouinlan J.R. Simplifying Decision Trees. Intemational Journal of Man-Machine Studies. 1987.3.221-248.
[21] Quinlan J.R. Induction of Decision Trees. Machine Learning. 1986.1.81-106
[22] Quinlan J.R. Improved Use of Continuous Attributes in C4.5. Journal of Artificial Intelligence Research. 1996.3.77-90.
[23] Dash M, Liu H. Feature Selection for Classification. Intelligent Data Analysis. 1997.1. 131-156
[24] Rissanen J、Wax M. Algorithm for Constructing Tree Structured Classifiers. US Patent. No.4719571.1988
[25] Mingers J. An Empirical Comparison of Pruning Methods for Decision Tree Induction,
Machine Learning. 1989.4. 227-243
[26] Oates T, Jensen D. The E fleets of Training Set Size On Decision Tree Complexity. Proceeding of the 14th International Conference On Machine Learning. 1997. 265-262
[27] Terri Moore, Carole Jesse、Riehard Kittler. An Overview and Evaluation of Decision Tree Methodology. ASA Quality and Productivity Conference. 2001.5.23-25
[28] Hess, Sidney W. Swinging on Branch of a Tree: Project Selection Application. Interfaces. 1993.12. 5-12
[29] Quinlan JR. C4.5: Programming for Machine Learning. CA. 1993
[30] Gunther Gediga, Ivo Duntseh. Statistical Tools for Rule Based Data Analysis. Synthesis of Intelligent Agent Systems From Experimental. ECAI98. 1998. 173-242
[31] Agrawal, Imielinski, Swami. Database Mining: A Performance Perspective. IEEE Transactions On Knowledge and Data Engineering. 1993.5. 914-925
[32] 刘同明．数据挖掘技术及其应用．北京：国防工业出版社，2001．9
[33] DanL Grecu, LeeA Becker. Coactive Learning for Distributed Data Mining. New York: 4th International Conference on Knowledge Discovery and Data Mining. 1998.7.
[34] Bhandari, Colet, Parker. Advanced Scout: Data Mining and Knowledge Discovery in NBA Data. Data Mining and Knowledge Discovery. 1997.1.121-125
[35] David Heckerman. Bayesian Networks for Data Mining. Data Mining and Knowledge Discovery. 1997.1.79-119
[36] 王国胤．Rough集理论与知识获取．西安：西安交通大学出版社．2001．5
[37] Z. Pawlak. Rough Sets. Int'I J. Compute & Science. 1982.10. 341-356
[38] Z. Pawlak. Rough Set Theory and Its Applications To Data Analysis. Cybernetics & Systems: An Int'I J. 1998. 661-688
[39] Ivo Dutseh, Gunter Gediga. The Rough Set Model for Data Analysis: Introduction and Overview. Preprint. http://www.infj.ulst.aciuk/～cccz23/papers/roughmod.html.1996.7
[40] Ivo Dutsch, Gunter Gediga. The Rough Set Engine GROBIAN. Seheol of Information and software Engineering. http://www.irfj.ulst.aciuk/～cccz23/papers/grobian.pdf
[41] Ziarko W, Shan. Rough Sets as a Methodology in Data Mining. 1998. 554-573
[42] 王钰、苗夺谦、周育健．关于Rough Set理论及应用的综述．模式识别与人工智能． 1996．12．337-344
[43] Ziarko W. Variable Precision Rough Set Model. Journal of Computer and System Sciences. 1993.46
[44] Ziarko W. AM ethod for Computing all Maximally General Rules In Attribute-value S.ystems. Computational Intelligence. 1993.2.2-13
[45] Bell D, Guan J. Computational Methods for Rough Classification and Discovery. Journal of ASIS 49. 1998.5.403-414
[46] Deogun J, Raghavan V, Sever H. Rough Set based classification methods and extended decision tables. In Proceedings of the International Workshop On Rough Sets and Soft
Computing. 1994. 302-309
[47] Sever H、Raghavan V. The Status of Research On Rough Sets For Knowledge Discovery In Databases. In Proceedings of the International Workshop On Rough Sets and Soft Computing. 1994. 325-332
[48] 常犁云、王国胤、吴渝．一种基于R0ugll Set理论的属性约简及规则提取方法．软件学报．1999．11．1206-1211
[49] 高毅龙．数据挖掘及其在工程诊断中的应用[博士学位论文]．2000．3
[50] Agrawal R, Imielienski T、Swami A. Mining Association Rules Between Sets of Items In Large Databases. Proc. Conf. On Management of Data. New York: ACM press. 1993. 10. 207-216
[51] Agrawal R, Mannila H、Srikant R. Fast Discovery of Association Rules. In: Fayyad et al. 1996. 307-328
[52] Hannu Toivonen. Sampling Large Databases for Association Rules. Proceeding of the 22nd VLDB Conference Mumbai(Bombay). India. 1996
[53] Zheng、Kohavi R、Mason L. Real World Performance of. Association Rule Algorithms. In: Proc. 7th Int. Conf On Knowledge Discovery and Data Mining(SIGKDD'01). 2001
[54] Mazlack J. Granulation of Quantitative Association Rules. International Journal of Fuzzy Systems. 2001.6. 400-406
[55] Brin S、Motwani R. Beyond Market Basket: Generalizing Association Rules To Correlations. In Proceeding of the ACM SIGMOD International Conference on Management of Data(ACM SIGMOD'97). 1997a. 265-276
[56] Bing Liu、Wynne Hsu、Yiming Ma. Integrating Classification and Association Rule Mining. KDD98. 1998.7.27-31
[57] Fukada T、Morimoto Y、Morishita S. Mining Optimized Association Rules for Numeric Attributes. In Proceeding of 15th ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems(PODS'96). 1996. 182-191
[58] Bing Liu、Wymae Hsu、Yiming Ma. Mining Association Rules with Multiple Minimum Supports. ACM SIGKDD International Conference On Knowledge Discovery & Data Mining (KDD'99). 1999.7
[59] Hipp J、Myka A、Wirth R. A New Algorithm for Faster Mining of Generalized Association Rules. In Proceedings of the 2nd European Symposium on Principles of Data Mining and Knowledge Discovery(PKDD'98). 1998.74-82
[60] Brin S、Motwani R. Dynamic Item Set Counting and Implication Rules for Market Basket Data. In Proceeding of the ACM SIGMOD International Conference on Management of Data(ACM SIGMOD'97). 1997b. 265-276
[61] Bing Liu、Wynne Hsu、Yiming Ma. Pruning and Summarizing the Discovered Associations. KDD'99. 1999.7
[62] 杨文献、任兴民．基于数据挖掘的复杂设备故障诊断方法研究．振动工程学报．
2000．10
[63] 李峰．数据库技术在汽轮机故障诊断系统中的应用．动力工程．第20卷．第1期．2000．2．549-553
[64] David Bennett．徐军等译．Visual C++5.0开发人员指南．北京：机械工业出版社．
[65] Richard C、Leinceker．陈冠民、王如亮等译．Visual C++5.0开发技术内幕．北京：机械工业出版社．