基于不完备信息的故障诊断知识获取技术研究
|
摘要
从包含大量矛盾、冗余、重复的设备运行状况征兆信息中进行诊断知识的获取,是整个智能故障诊断过程的核心,也是建造智能诊断系统的重点和难点。在实际的诊断知识获取过程中,常常由于所得到的知识内容不完整、结构不完备而无法获取到完备的故障诊断知识。目前故障广泛应用的各种故障诊断知识获取方法多是针对诊断信息完备的情况,导致在故障诊断的知识获取中无法充分利用不完备诊断信息,降低了故障诊断知识获取的有效性和诊断结果的正确率。
本文结合故障诊断问题的特点,总结了故障诊断中不完备信息的产生原因,研究了不完备诊断信息的语义解释,并给出不完备故障诊断信息系统的定义。结合现有的不完备信息系统的扩充关系模型,建立了一种适合故障诊断不完备决策系统的关系模型,用来描述含有不完备信息的诊断知识之间的关系。
在此基础上,给出完备与不完备故障诊断决策流图的定义,将不完备故障诊断决策流图中的节点定义为由征兆属性和其取值联合描述的二元组;提出将不完备故障诊断决策流图中的分支区分为有向和无向相结合的描述方式,并给出了采用流经分支的对象集合作为分支描述参数的思路,建立了计算流经各种不同分支上流量的公式。由此,建立了一种适合于包含不完备诊断信息的故障诊断知识的图形化方法。
本文分析了不完备故障诊断决策流图中隐含的决策规则,并采用置信度和覆盖度的概念作为决策规则的评价指标。针对不完备故障诊断决策流图,分别进行了层次约简和节点约简。约简后得到的不完备故障诊断决策流图不损失原有的故障诊断数据信息且不含冗余的信息,以最简的方式表达了征兆属性和决策属性间的关系。
故障诊断应用软件可以完成基于流图的完备或不完备的故障诊断知识获取,可视化整个知识发现过程。本文对应用软件的设计思想与开发方案进行了简要的阐述。另外,本文以船用柴油机故障诊断为例进行了研究,进一步说明了应用软件的使用方法和有效性,验证了本文研究成果的正确性。
It’s a key for a intelligent fault diagnosis process to acquire effectiveknowledge from a great deal of symptomless information about equipment functionstatus. Repetitious. Most of the fault diagnosis means, which are widely used atpresent, are for the complete fault dianosis system. While incomplete informationabout fault object is usually gained because practice is restricted by objectiveenvironment and factor. So a new method aimed at incomplete fault dianosis systemis needed to make full use of the information, and to boost the validity of theknowledge acquisition for fault dianosis.
Main contents of this paper are as follows:
According as the character of the fault diagnosis problems, this paper hassummarized the reasons why the incomplete information exsisted in the faultdiagnosis, has investigated the explanation for the incomplete diagnodsisinformation, has definated the incomplete fault diagnosis information system.Combining with the exsisted relation former for the incomplete information, thepaper has established advanced-restricted tolerance relation. This relation is fit forthe incompleted fault diagnosis information system to describe the connectionamong the incomplete information.
The paper has definated the complete and incomplete fault diagnosis decisionflow graphs, has introduced a binary-group including symptom attribute and itsvalue to definate circumscribe the notes in the incomplete fault diagnosis decisionflow graphs, has distinguished between the directing embranchments and theundirecting embranchments, has adopted the set of the objects by theembranchment to characterize a embranchment and definated the flux formula of aembranchment. On the basis of all above, a graphics method is introduced toexpress the incomplete diagnosis knowledge.
The paper has analysised the connotative decision rules in incomplete faultdiagnosis decision flow graphs, and has imported the concept of certainty andcoverage factors as the judgement for decision rules. Tier reduction and notereduction are two steps for the structure reduction, the paper has explained how toreduct the tiers and notes respectively. The reducted incomplete fault diagnosisdecision-making flow graphs doesn’t lost any intrinsic diagnosis information anddoesn’t contain any redundant data, it delivers oneself of the relations between thesymptom attributes and the decision attributes. Otherwise, fault diagnosis of diesel engines for ships has been taken as an example to introduce how to use the softwareand how resultful the software is. The example has validated the production of thepaper likewise.
本文结合故障诊断问题的特点,总结了故障诊断中不完备信息的产生原因,研究了不完备诊断信息的语义解释,并给出不完备故障诊断信息系统的定义。结合现有的不完备信息系统的扩充关系模型,建立了一种适合故障诊断不完备决策系统的关系模型,用来描述含有不完备信息的诊断知识之间的关系。
在此基础上,给出完备与不完备故障诊断决策流图的定义,将不完备故障诊断决策流图中的节点定义为由征兆属性和其取值联合描述的二元组;提出将不完备故障诊断决策流图中的分支区分为有向和无向相结合的描述方式,并给出了采用流经分支的对象集合作为分支描述参数的思路,建立了计算流经各种不同分支上流量的公式。由此,建立了一种适合于包含不完备诊断信息的故障诊断知识的图形化方法。
本文分析了不完备故障诊断决策流图中隐含的决策规则,并采用置信度和覆盖度的概念作为决策规则的评价指标。针对不完备故障诊断决策流图,分别进行了层次约简和节点约简。约简后得到的不完备故障诊断决策流图不损失原有的故障诊断数据信息且不含冗余的信息,以最简的方式表达了征兆属性和决策属性间的关系。
故障诊断应用软件可以完成基于流图的完备或不完备的故障诊断知识获取,可视化整个知识发现过程。本文对应用软件的设计思想与开发方案进行了简要的阐述。另外,本文以船用柴油机故障诊断为例进行了研究,进一步说明了应用软件的使用方法和有效性,验证了本文研究成果的正确性。
It’s a key for a intelligent fault diagnosis process to acquire effectiveknowledge from a great deal of symptomless information about equipment functionstatus. Repetitious. Most of the fault diagnosis means, which are widely used atpresent, are for the complete fault dianosis system. While incomplete informationabout fault object is usually gained because practice is restricted by objectiveenvironment and factor. So a new method aimed at incomplete fault dianosis systemis needed to make full use of the information, and to boost the validity of theknowledge acquisition for fault dianosis.
Main contents of this paper are as follows:
According as the character of the fault diagnosis problems, this paper hassummarized the reasons why the incomplete information exsisted in the faultdiagnosis, has investigated the explanation for the incomplete diagnodsisinformation, has definated the incomplete fault diagnosis information system.Combining with the exsisted relation former for the incomplete information, thepaper has established advanced-restricted tolerance relation. This relation is fit forthe incompleted fault diagnosis information system to describe the connectionamong the incomplete information.
The paper has definated the complete and incomplete fault diagnosis decisionflow graphs, has introduced a binary-group including symptom attribute and itsvalue to definate circumscribe the notes in the incomplete fault diagnosis decisionflow graphs, has distinguished between the directing embranchments and theundirecting embranchments, has adopted the set of the objects by theembranchment to characterize a embranchment and definated the flux formula of aembranchment. On the basis of all above, a graphics method is introduced toexpress the incomplete diagnosis knowledge.
The paper has analysised the connotative decision rules in incomplete faultdiagnosis decision flow graphs, and has imported the concept of certainty andcoverage factors as the judgement for decision rules. Tier reduction and notereduction are two steps for the structure reduction, the paper has explained how toreduct the tiers and notes respectively. The reducted incomplete fault diagnosisdecision-making flow graphs doesn’t lost any intrinsic diagnosis information anddoesn’t contain any redundant data, it delivers oneself of the relations between thesymptom attributes and the decision attributes. Otherwise, fault diagnosis of diesel engines for ships has been taken as an example to introduce how to use the softwareand how resultful the software is. The example has validated the production of thepaper likewise.
引文
1.黄文虎,夏松波,刘瑞岩等.设备故障诊断原理、技术及应用.科学出版社,1996:1~15
2.吴今培,肖健华.智能故障诊断与专家系统.科学出版社, 1997:1~77
3.王道平,张义忠.故障智能诊断系统的理论与方法.冶金工业出版社,2001:1~62
4.刘宜平.基于智能信息处理技术的故障诊断方法研究.哈尔滨工业大学博士学位论文. 1999:1~18
5.屈梁生,孟健.机械故障诊断技术与当代前沿科学.科学出版社, 1997:1~77
6. Li Jing Rong, Khoo Li Pheng, Tor, Shu Beng. RMINE: A rough set based datamining prototype for the reasoning of incomplete data in condition-based faultdiagnosis. Journal of Intelligent Manufacturing, 2006, 17(1): 163-176
7. A. Isabel, F. Gabriel. Approach to the evaluation of a Diagnostic System withincomplete information. IEEE Symposium on Emerging Technologies andFactory Automation, 1999, 2: 963-969
8.罗党.灰色决策问题的分析方法研究.南京航空航天大学博士学位论文.2005
9.李建波,赵继印,郑蕊蕊.基于综合关联度分析的电力变压器故障诊断.吉林大学学报(信息科学版). 2008,26(1):62~68
10. A.Zadeh. Fuzzy Sets. Information and Control. 1965, 8(3):338~353
11. Wang Huaqing, Chen Peng. Fault diagnosis for a rolling bearing used in areciprocating machine by adaptive filtering technique and fuzzy neural network.WSEAS Transactions on Systems, 2008,7(1): 1-6
12. Dong Lixin, Xiao Dengming, Liang Yishan, Liu Yilu. Rough set and fuzzywavelet neural network integrated with least square weighted fusion algorithmbased fault diagnosis research for power transformers. Electric Power SystemsResearch, 2008,78(1):129-136
13. Chen J., Roberts C., Weston P. Fault detection and diagnosis for railway trackcircuits using neuro-fuzzy systems. Control Engineering Practice,2008,16(5):585-596
14. Luo Xu; Kezunovic Mladen. Implementing fuzzy reasoning Petri-nets for faultsection estimation. IEEE Transactions on Power Delivery, 2008,23(2):676-685
15. Z.Pawlak. Rough Sets. International Journal of Computer and InformationSciences. 1982,11(5):341~356
16. Z.Pawlak. Rough Sets: Theoretical Aspects of Reasoning about Data. Boston:Kluwer Academic Publishers, 1991:1~62
17. Z.Pawlak et al. Rough Sets. Communications of the ACM. 1995,38(11): 89~95
18. Chen Xianghui, Zhu Shanjun, Ji Yindong. Entropy Based Uncertainty Measuresfor Classification Rules with Inconsistency Tolerance. Proceedings of the IEEEInternational Conference on Systems, Man and Cybernetics. 2000,4:2816~2821
19. G.Y. Wang, Y. Wu, F. Liu. Generating Rules and Reasoning underInconsistencies. 26th Annual Conference of the IEEE Electronics SocietyIECON 2000. Nagoya. 2000, 4: 2536~2541
20. M. Kryszkiewicz. Comparative Study of Alternative Types of KnowledgeReduction in Inconsistent Systems. International Journal of Intelligent Systems.2001, 16(1):105~120
21. J. W. Grzymala-Busse, W. J. Grzymala-Busse, L. K. Goodwin. A Comparisonof Three Closest Fit Approaches to Missing Attribute Values in Preterm BirthData. International Journal of Intelligent Systems. 2002, 17(2):125~134
22. J.W. Grzymala-Busse, W.J. Grzymala-Busse, L.K. Goodwin. Coping withMissing Attribute Values based on Closest Fit in Preterm Birth Data: A RoughSet Approach. Computational Intelligence. 2001,17(3):425~434
23. J.W. Grzymala-Busse, Ming Hu. A Comparison of Several Approaches toMissing Attribute Values in Data Mining. Proceedings of the SecondInternational Conference on Rough Sets and Current Trends in ComputingRSCTC'2000. Banff, Canada. 2000:340~347
24. N. Zhong, J. Z. Dong, S. Ohsuga, T. Y. Lin. An Incremental, ProbabilisticRough Set Approach to Rule Discovery. Proceedings of the 1998 IEEEInternational Conference on Fuzzy Systems, Anchorage, AK, USA.1998,2:933~938
25. Hong Tzung-Pei, Tseng Li-Huei, Wang Shyue-Liang. Learning Rules fromIncomplete Training Examples by Rough Sets. Expert Systems withApplications. 2002, 22(4):285~293
26. Hong Tzung-Pei, Tseng Li-Huei, Chien Been-Chian. Learning Fuzzy Rulesfrom Incomplete Quantitative Data by Rough Sets. IEEE InternationalConference on Plasma Science. 2002, 2:1438~1443
27. P.J. Lingras, Y.Y. Yao. Data Mining using Extensions of the Rough Set Model.Journal of the American Society for Information Science. 1998,49(5):415~422
28. Y. Leung, Li Deyu. Maximal Consistent Block Technique for Rule Acquisitionin Incomplete Information Systems. Information Sciences. 2003,153(SUPP):85~106
29. X.H. Hu, N. Cercone. Discovering Maximal Generalized Decision Rulesthrough Horizontal and Vertical Data Reduction. Computational Intelligence.2001,17(4):685~702
30.胡近海,谢寿生,汪诚,李应红,杨帆.基于粗糙核Fisher鉴别分析的特征提取及其在发动机故障诊断中的应用.航空动力学报, 2008,23(7):1346-1352
31.熊浩,李卫国,畅广辉,郭惠敏.模糊粗糙集理论在变压器故障诊断中的应用.中国电机工程学报, 2008,28(7):141-147
32.王英英,罗毅,涂光瑜.基于粗糙集与决策树的配电网故障诊断方法.高电压技术, 2008,34(4):794-798
33.陈果.粗糙集-遗传算法-神经网络集成分类器及其在转子故障诊断中的应用研究.中国机械工程, 2008,19(1):85-90
34. Z. Pawlak. Theorize with data using rough sets. Proceedings - IEEE ComputerSociety's International Computer Software and Applications Conference. 2002:1125-1128
35. Z. Pawlak. Flow Graphs and Intelligent Data Analysis. FundamentaInformaticae, 2005, 64(1-4): 369-377
36. Z. Pawlak. Decision trees and flow graphs. Lecture Notes in Computer Science(Rough Sets and Current Trends in Computing - 5th International Conference),2006: 1-11
37.黄文涛,王伟杰,赵学增.从不完备数据中获取诊断规则的粗糙集方法.电力系统自动化, 2005,29(14): 49-54
38.黄文涛,王伟杰,赵学增,孟庆鑫.汽轮发电机组故障诊断的“规则+例外”知识获取模型.电力自动化设备. 2007,27(12):27~31
39. Huang Wentao, Wang Weijie, Zhao Xuezeng. Rule extracting method inincomplete decision table for fault diagnosis based on discernibility matrixprimitive. Proceeding of the 3rd International Conference on Rough Sets andKnowledge Tecnology, RSKT 2008. 2008:755-762
40. Huang Wentao, Wang Weijie, Meng Qingcin. A Discernibility MaxtrixPrimitive Based Rule Extraction Method for Wear Pattern Recognition ofDiesel Engine from Incomplete Data. Presentation on the 7th World Congresson Intelligent and Automation(WCICA’08), 2008: 1956-1960
41.黄文涛,王伟杰,孟庆鑫.基于粗糙集理论的电力变压器故障诊断方法. 2008中国控制与决策会议(20th CCDC), 2008: 468-472
42. M. Kryszkiewicz. Rough Set Approach to Incomplete Information Systems.Information Sciences. 1998,112(1-4):39~49
43. M. Kryszkiewicz. Rules in Incomplete Information Systems. InformationSciences. 1999,113(3-4):271~292
44. M. Kryszkiewicz. Properties of Incomplete Information Systems in theFramework of Rough Sets. L. Polkowski, A. Skowron. Rough Sets inKnowledge Discovery 1. Physica-Verlag, 1998:422~450
45. R. Slowinski, D. Vanderpooten. A Generalized Definition of RoughApproximations Based on Similarity. IEEE Transactions on Knowledge andData Engineering. 2000, 12(2):331~336
2.吴今培,肖健华.智能故障诊断与专家系统.科学出版社, 1997:1~77
3.王道平,张义忠.故障智能诊断系统的理论与方法.冶金工业出版社,2001:1~62
4.刘宜平.基于智能信息处理技术的故障诊断方法研究.哈尔滨工业大学博士学位论文. 1999:1~18
5.屈梁生,孟健.机械故障诊断技术与当代前沿科学.科学出版社, 1997:1~77
6. Li Jing Rong, Khoo Li Pheng, Tor, Shu Beng. RMINE: A rough set based datamining prototype for the reasoning of incomplete data in condition-based faultdiagnosis. Journal of Intelligent Manufacturing, 2006, 17(1): 163-176
7. A. Isabel, F. Gabriel. Approach to the evaluation of a Diagnostic System withincomplete information. IEEE Symposium on Emerging Technologies andFactory Automation, 1999, 2: 963-969
8.罗党.灰色决策问题的分析方法研究.南京航空航天大学博士学位论文.2005
9.李建波,赵继印,郑蕊蕊.基于综合关联度分析的电力变压器故障诊断.吉林大学学报(信息科学版). 2008,26(1):62~68
10. A.Zadeh. Fuzzy Sets. Information and Control. 1965, 8(3):338~353
11. Wang Huaqing, Chen Peng. Fault diagnosis for a rolling bearing used in areciprocating machine by adaptive filtering technique and fuzzy neural network.WSEAS Transactions on Systems, 2008,7(1): 1-6
12. Dong Lixin, Xiao Dengming, Liang Yishan, Liu Yilu. Rough set and fuzzywavelet neural network integrated with least square weighted fusion algorithmbased fault diagnosis research for power transformers. Electric Power SystemsResearch, 2008,78(1):129-136
13. Chen J., Roberts C., Weston P. Fault detection and diagnosis for railway trackcircuits using neuro-fuzzy systems. Control Engineering Practice,2008,16(5):585-596
14. Luo Xu; Kezunovic Mladen. Implementing fuzzy reasoning Petri-nets for faultsection estimation. IEEE Transactions on Power Delivery, 2008,23(2):676-685
15. Z.Pawlak. Rough Sets. International Journal of Computer and InformationSciences. 1982,11(5):341~356
16. Z.Pawlak. Rough Sets: Theoretical Aspects of Reasoning about Data. Boston:Kluwer Academic Publishers, 1991:1~62
17. Z.Pawlak et al. Rough Sets. Communications of the ACM. 1995,38(11): 89~95
18. Chen Xianghui, Zhu Shanjun, Ji Yindong. Entropy Based Uncertainty Measuresfor Classification Rules with Inconsistency Tolerance. Proceedings of the IEEEInternational Conference on Systems, Man and Cybernetics. 2000,4:2816~2821
19. G.Y. Wang, Y. Wu, F. Liu. Generating Rules and Reasoning underInconsistencies. 26th Annual Conference of the IEEE Electronics SocietyIECON 2000. Nagoya. 2000, 4: 2536~2541
20. M. Kryszkiewicz. Comparative Study of Alternative Types of KnowledgeReduction in Inconsistent Systems. International Journal of Intelligent Systems.2001, 16(1):105~120
21. J. W. Grzymala-Busse, W. J. Grzymala-Busse, L. K. Goodwin. A Comparisonof Three Closest Fit Approaches to Missing Attribute Values in Preterm BirthData. International Journal of Intelligent Systems. 2002, 17(2):125~134
22. J.W. Grzymala-Busse, W.J. Grzymala-Busse, L.K. Goodwin. Coping withMissing Attribute Values based on Closest Fit in Preterm Birth Data: A RoughSet Approach. Computational Intelligence. 2001,17(3):425~434
23. J.W. Grzymala-Busse, Ming Hu. A Comparison of Several Approaches toMissing Attribute Values in Data Mining. Proceedings of the SecondInternational Conference on Rough Sets and Current Trends in ComputingRSCTC'2000. Banff, Canada. 2000:340~347
24. N. Zhong, J. Z. Dong, S. Ohsuga, T. Y. Lin. An Incremental, ProbabilisticRough Set Approach to Rule Discovery. Proceedings of the 1998 IEEEInternational Conference on Fuzzy Systems, Anchorage, AK, USA.1998,2:933~938
25. Hong Tzung-Pei, Tseng Li-Huei, Wang Shyue-Liang. Learning Rules fromIncomplete Training Examples by Rough Sets. Expert Systems withApplications. 2002, 22(4):285~293
26. Hong Tzung-Pei, Tseng Li-Huei, Chien Been-Chian. Learning Fuzzy Rulesfrom Incomplete Quantitative Data by Rough Sets. IEEE InternationalConference on Plasma Science. 2002, 2:1438~1443
27. P.J. Lingras, Y.Y. Yao. Data Mining using Extensions of the Rough Set Model.Journal of the American Society for Information Science. 1998,49(5):415~422
28. Y. Leung, Li Deyu. Maximal Consistent Block Technique for Rule Acquisitionin Incomplete Information Systems. Information Sciences. 2003,153(SUPP):85~106
29. X.H. Hu, N. Cercone. Discovering Maximal Generalized Decision Rulesthrough Horizontal and Vertical Data Reduction. Computational Intelligence.2001,17(4):685~702
30.胡近海,谢寿生,汪诚,李应红,杨帆.基于粗糙核Fisher鉴别分析的特征提取及其在发动机故障诊断中的应用.航空动力学报, 2008,23(7):1346-1352
31.熊浩,李卫国,畅广辉,郭惠敏.模糊粗糙集理论在变压器故障诊断中的应用.中国电机工程学报, 2008,28(7):141-147
32.王英英,罗毅,涂光瑜.基于粗糙集与决策树的配电网故障诊断方法.高电压技术, 2008,34(4):794-798
33.陈果.粗糙集-遗传算法-神经网络集成分类器及其在转子故障诊断中的应用研究.中国机械工程, 2008,19(1):85-90
34. Z. Pawlak. Theorize with data using rough sets. Proceedings - IEEE ComputerSociety's International Computer Software and Applications Conference. 2002:1125-1128
35. Z. Pawlak. Flow Graphs and Intelligent Data Analysis. FundamentaInformaticae, 2005, 64(1-4): 369-377
36. Z. Pawlak. Decision trees and flow graphs. Lecture Notes in Computer Science(Rough Sets and Current Trends in Computing - 5th International Conference),2006: 1-11
37.黄文涛,王伟杰,赵学增.从不完备数据中获取诊断规则的粗糙集方法.电力系统自动化, 2005,29(14): 49-54
38.黄文涛,王伟杰,赵学增,孟庆鑫.汽轮发电机组故障诊断的“规则+例外”知识获取模型.电力自动化设备. 2007,27(12):27~31
39. Huang Wentao, Wang Weijie, Zhao Xuezeng. Rule extracting method inincomplete decision table for fault diagnosis based on discernibility matrixprimitive. Proceeding of the 3rd International Conference on Rough Sets andKnowledge Tecnology, RSKT 2008. 2008:755-762
40. Huang Wentao, Wang Weijie, Meng Qingcin. A Discernibility MaxtrixPrimitive Based Rule Extraction Method for Wear Pattern Recognition ofDiesel Engine from Incomplete Data. Presentation on the 7th World Congresson Intelligent and Automation(WCICA’08), 2008: 1956-1960
41.黄文涛,王伟杰,孟庆鑫.基于粗糙集理论的电力变压器故障诊断方法. 2008中国控制与决策会议(20th CCDC), 2008: 468-472
42. M. Kryszkiewicz. Rough Set Approach to Incomplete Information Systems.Information Sciences. 1998,112(1-4):39~49
43. M. Kryszkiewicz. Rules in Incomplete Information Systems. InformationSciences. 1999,113(3-4):271~292
44. M. Kryszkiewicz. Properties of Incomplete Information Systems in theFramework of Rough Sets. L. Polkowski, A. Skowron. Rough Sets inKnowledge Discovery 1. Physica-Verlag, 1998:422~450
45. R. Slowinski, D. Vanderpooten. A Generalized Definition of RoughApproximations Based on Similarity. IEEE Transactions on Knowledge andData Engineering. 2000, 12(2):331~336