Bayesian网络在制动系统故障诊断中的应用及系统开发
|
摘要
制动系统是汽车设备中至关重要的运行设备,其运行状态直接关系到人员的安全,因此,在制动系统运行状态检测的基础上展开故障诊断就显得尤为必要。但是,由于制动系统中存在很多错综复杂的相互关系,并存在大量的不确定因素及不确定信息,使得故障诊断较为困难。针对以上问题,本文对贝叶斯网络在液压制动系统故障诊断中的应用进行了研究。
贝叶斯网络是目前不确定性知识表达和推理领域最有效的理论模型之一,适用于不确定性和概率推理的知识表达和推理。它是一种基于网络结构的有向图解描述,能进行双向并行推理,并能综合先验信息和样本信息,使得推理结果更为准确可信。因此,贝叶斯网络在故障诊断领域中的应用具有重要意义。
本文以液压制动系统的故障诊断为研究对象,在对液压制动系统故障模式及原因分析的基础之上,提出了用贝叶斯网络来解决液压制动系统故障诊断的方法。根据多专家提供的规则进行贝叶斯网络结构学习,建立了基于贝叶斯网络的液压制动系统故障诊断分层结构模型,对模型的知识表达、建造方法进行了深入研究,同时也对贝叶斯网络基于团树传播的精确推理方法进行了论述。然后设计了贝叶斯网络故障诊断系统,并用c#语言实现了从建造贝叶斯网络结构,到进行诊断推理得出诊断结论整个过程。最后给出了一个例子来分析如何利用故障诊断系统进行故障诊断。
经实验数据分析表明：本文的故障诊断系统诊断准确率高于模糊逻辑方法诊断准确率9.96个百分点,有效地解决了故障诊断中存在的不确定性问题,提高了诊断的准确率,从而验证了本文的故障诊断模型的有效性和具体的应用价值。
The braking system is one of the important operation equipment in the automotive equipment and its running state directly relates to the safety of personnel, Therefore, it is particularly necessary to spread fault diagnosis, which is based on the check-up of running state in the braking system. But because of a lot of anfractuous correlations in the braking system and the existing of many uncertainty factors and much uncertainty information, it is very difficult to carry through fault diagnosis.In view of the above problems, the author made a research on applications of Bayesian network in the fault diagnosis of hydraulic braking system.
The Bayesian network is one of the most efficient theoretical models in the current uncertainty of knowledge and inference areas. It can be used to the inference and knowledge expression of uncertainty and probability inference. As a exist-way diagram description based on net construction, it can start a two-way parallel inference, synthesize pre-tested and sample information and make the result of inference more accurate and credible. Therefore, Bayesian network has a deep impact on applications of fault diagnosis.
This paper based on the fault diagnosis of hydraulic braking system for the study object, and on the basis of the analysis of the hydraulic braking system fault mode and the reasons put forward a method by Bayesian network to solve the fault diagnosis of hydraulic braking system. According to many rules provided by experts, the author learned the network structure of Bayesian,built the fault diagnosis of hydraulic braking system hierarchical structure model based on Bayesian network,carried out in-depth research about the knowledge expression and construction method, and discussed the accurate inference approaches based on clique tree. Then the author designed a Bayesian network fault diagnosis system, and using c# language realized the process from the construction of Bayesian network to a diagnostic inference drawn conclusions. Finally, the author gave an example of how to use fault diagnosis system for fault diagnosis.
The analysis of the experimental data showed:In this paper, the diagnostic accuracy rate of the fault diagnosis system is 9.96 percentage points higher than the fuzzy logic method diagnostic accuracy rate, the uncertain problems of fault diagnosis were availably solved. The paper enhanced the diagnostic accuracy rate, as a result, proved the effectiveness of fault diagnosis model and idiographic application value.
贝叶斯网络是目前不确定性知识表达和推理领域最有效的理论模型之一,适用于不确定性和概率推理的知识表达和推理。它是一种基于网络结构的有向图解描述,能进行双向并行推理,并能综合先验信息和样本信息,使得推理结果更为准确可信。因此,贝叶斯网络在故障诊断领域中的应用具有重要意义。
本文以液压制动系统的故障诊断为研究对象,在对液压制动系统故障模式及原因分析的基础之上,提出了用贝叶斯网络来解决液压制动系统故障诊断的方法。根据多专家提供的规则进行贝叶斯网络结构学习,建立了基于贝叶斯网络的液压制动系统故障诊断分层结构模型,对模型的知识表达、建造方法进行了深入研究,同时也对贝叶斯网络基于团树传播的精确推理方法进行了论述。然后设计了贝叶斯网络故障诊断系统,并用c#语言实现了从建造贝叶斯网络结构,到进行诊断推理得出诊断结论整个过程。最后给出了一个例子来分析如何利用故障诊断系统进行故障诊断。
经实验数据分析表明：本文的故障诊断系统诊断准确率高于模糊逻辑方法诊断准确率9.96个百分点,有效地解决了故障诊断中存在的不确定性问题,提高了诊断的准确率,从而验证了本文的故障诊断模型的有效性和具体的应用价值。
The braking system is one of the important operation equipment in the automotive equipment and its running state directly relates to the safety of personnel, Therefore, it is particularly necessary to spread fault diagnosis, which is based on the check-up of running state in the braking system. But because of a lot of anfractuous correlations in the braking system and the existing of many uncertainty factors and much uncertainty information, it is very difficult to carry through fault diagnosis.In view of the above problems, the author made a research on applications of Bayesian network in the fault diagnosis of hydraulic braking system.
The Bayesian network is one of the most efficient theoretical models in the current uncertainty of knowledge and inference areas. It can be used to the inference and knowledge expression of uncertainty and probability inference. As a exist-way diagram description based on net construction, it can start a two-way parallel inference, synthesize pre-tested and sample information and make the result of inference more accurate and credible. Therefore, Bayesian network has a deep impact on applications of fault diagnosis.
This paper based on the fault diagnosis of hydraulic braking system for the study object, and on the basis of the analysis of the hydraulic braking system fault mode and the reasons put forward a method by Bayesian network to solve the fault diagnosis of hydraulic braking system. According to many rules provided by experts, the author learned the network structure of Bayesian,built the fault diagnosis of hydraulic braking system hierarchical structure model based on Bayesian network,carried out in-depth research about the knowledge expression and construction method, and discussed the accurate inference approaches based on clique tree. Then the author designed a Bayesian network fault diagnosis system, and using c# language realized the process from the construction of Bayesian network to a diagnostic inference drawn conclusions. Finally, the author gave an example of how to use fault diagnosis system for fault diagnosis.
The analysis of the experimental data showed:In this paper, the diagnostic accuracy rate of the fault diagnosis system is 9.96 percentage points higher than the fuzzy logic method diagnostic accuracy rate, the uncertain problems of fault diagnosis were availably solved. The paper enhanced the diagnostic accuracy rate, as a result, proved the effectiveness of fault diagnosis model and idiographic application value.
引文
1.杨为民.可靠性、维修性、保障性总论[M],北京：国防工业出版社,1995,15-136.
2.柳卫东.汽车制动系统FTA法的故障诊断研究[D],西安：西北工业大学,2007.
3.傅军,贺炜,阎建国等.贝叶斯网络在柴油机动力装置故障诊断中的应用[J],上海海运学院学报,2001,22(3)：68-72.
4.张宏辉,唐锡宽.贝叶斯推理网络在大型旋转机械故障诊断中的应用[J],机械科学与技术,1996,25(2)：43-46.
5.徐宾刚,屈梁生,陶肖明.转子贝叶斯诊断网络的研究[J],机械工程学报,2004,40(1)：68-72.
6.华斌,周建中,喻菁.贝叶斯网络在水电机组故障诊断中的应用研究[J],华北电力大学学报,2004,31(5)：33-36.
7.张德利,韩璞,焦嵩鸣.贝叶斯网络在汽轮机故障诊断中的应用[J],微计算机信息,2007,23(10)：144-146.
8.Przytula K,Hagen F,Yung K. Bayesian Networks for Satellite Payload Testing[D], Proceeding of the Forty-Fourth SP IE,Denver,1999.
9.党明瑞,董宇翔.用Bayesian网络对光纤接入网供电系统故障的监控[J],半导体光电,2000,21(4)：286-290.
10.傅军,贺炜,阎建国等.贝叶斯网络在柴油机动力装置故障诊断中的应用[J],上海海运学院学报,2001,22(3)：68-72.
11.张宏辉,唐锡宽.贝叶斯推理网络在大型旋转机械故障诊断中的应用[J],机械科学与技术,1996,25(2)：43-46.
12.Heckerman D,Rommelse K.Decisiontheoretic T roubleshooting[J],Communications of the ACM,1995,38(3):49-57.
13.Beiser J A,R igdon S E. Bayes P rediction for the Number of Failures of A Repairable System IEEE Transactions on Reliability[J],1997,46(2):320-326.
14.张连文,郭海鹏.贝叶斯网引论[M],北京：科学出版社,2006,3-138.
15.刘志强.贝叶斯网络与认知图[J],自动化学报,2001,27(4)：52-55.
16. Dawid A P. Statistical theory,the prequential approach[J],Journal of the Royal Statistical Society,1984,147(2):278-292.
17.刘启元,张聪等.信度网近似推理算法(上)[J],计算机科学,2001,28(1)：70-73.
18.刘伟娜,霍利民,张立国.贝叶斯网络精确推理算法的研究[J],微计算机信息,2006,22(3)：92-94.
19.阎平凡.不精确推理中的信度网络方法[J],计算机科学,1989,5,53-56.
20.刘振,史建国,代进进.引入证据后的贝叶斯网络信息传递方法[J],计算机应用,2007,27：85-86.
21.俞奎.贝叶斯网络建模及推理算法研究[D],合肥：合肥工业大学,2007.
22.董立岩,苑森淼,刘光远.基于预测能力的连续贝叶斯网络结构学习[J],计算机工程与应用,2007,43(9)：23-25.
23.Chickering D M.Learning Bayesian Networks is NP-complete[J],learning from data:AI and Statistics,New York:Springer,1996:121-130.
24. Heckerman D.A tutorial on learning with Bayesian network (Technical Report MSR TR-95-06)[M],Redmond:Microsoft Research,1995,42-49.
25.Cheng J.Learning Bayesian networks from date[D],An information theory based approach,Unirersity of lster U K,1998:53-67.
26.王双成,林士敏等.贝叶斯网络结构学习分析[J],计算机科学,2002,27(10)：77-79.
27.王双成,林士敏等.贝叶斯学习、贝叶斯网络与数据采掘[J],计算机科学,2000,27(10)：69-72.
28.王玮,蔡莲红.贝叶斯网络拓扑结构确定方法的研究[J],小型微型计算机系统,2002,23(4)：435-437.
29.林士敏,王双成,陆玉昌.Bayesian方法的计算学习机制和问题求解[J],清华大学学报(自然科学版),2000,40(9)：61-64.
30.余志生.汽车理论[M],北京：机械工业出版社,2000,10-89.
31.齐晓杰.制动系统[M],北京：化学工业出版社,2005,4-108.
32.Shenoy P P,Shafer G.Axioms for probability and belief function propagation[J], Uncertainty in artificial intelligence,Amsterdam:Elsevier Science Publishers,1990, 169-198.
33.J,Pearl.Probabilistic reasoning in intelligent system:Networks of plausible inference[M], San Mateo,CA:Morgan Kaufmann Publishers,1988,44-48.
34. Xiao-dan Zhang,Hai Zhao.A General Situation Assessment Model Based on Rule Extraction[J],Proceedings of the 7th International Conference on Information Fusion, Stockholm,Sweden,Jun.2004,pp:1038-1043.
35.王清印,崔援民,赵秀恒等.预测与决策的不确定性数学模型[M],北京：冶金工业出版社,2001.
36.王道平,冯振声等.故障诊断专家系统中不确定性问题[J],计算机工程与应用,1998,34(12)：33-35.
37.李海成.一种基于规则库的贝叶斯网络开发器的设计与实现[J],陕西科技大学学报,2005,23：93-96.
38.吴泰平.基于故障树分析法与模糊理论的汽车制动系统故障诊断研究[D],广西：广西大学,2004.
39.彭亚敏.模糊逻辑在汽车故障诊断技术中的应用[J],计算机与数字工程,2002,10(4)：24-29.
40.刘喜红,吴康雄.模糊逻辑系统在机械故障诊断中的应用[J],设计与研究,2006,33(2)：15-17.
41.操召发.基于模糊逻辑推理故障智能诊断系统及其仿真研究[D],上海：上海海事大学,2004.
2.柳卫东.汽车制动系统FTA法的故障诊断研究[D],西安：西北工业大学,2007.
3.傅军,贺炜,阎建国等.贝叶斯网络在柴油机动力装置故障诊断中的应用[J],上海海运学院学报,2001,22(3)：68-72.
4.张宏辉,唐锡宽.贝叶斯推理网络在大型旋转机械故障诊断中的应用[J],机械科学与技术,1996,25(2)：43-46.
5.徐宾刚,屈梁生,陶肖明.转子贝叶斯诊断网络的研究[J],机械工程学报,2004,40(1)：68-72.
6.华斌,周建中,喻菁.贝叶斯网络在水电机组故障诊断中的应用研究[J],华北电力大学学报,2004,31(5)：33-36.
7.张德利,韩璞,焦嵩鸣.贝叶斯网络在汽轮机故障诊断中的应用[J],微计算机信息,2007,23(10)：144-146.
8.Przytula K,Hagen F,Yung K. Bayesian Networks for Satellite Payload Testing[D], Proceeding of the Forty-Fourth SP IE,Denver,1999.
9.党明瑞,董宇翔.用Bayesian网络对光纤接入网供电系统故障的监控[J],半导体光电,2000,21(4)：286-290.
10.傅军,贺炜,阎建国等.贝叶斯网络在柴油机动力装置故障诊断中的应用[J],上海海运学院学报,2001,22(3)：68-72.
11.张宏辉,唐锡宽.贝叶斯推理网络在大型旋转机械故障诊断中的应用[J],机械科学与技术,1996,25(2)：43-46.
12.Heckerman D,Rommelse K.Decisiontheoretic T roubleshooting[J],Communications of the ACM,1995,38(3):49-57.
13.Beiser J A,R igdon S E. Bayes P rediction for the Number of Failures of A Repairable System IEEE Transactions on Reliability[J],1997,46(2):320-326.
14.张连文,郭海鹏.贝叶斯网引论[M],北京：科学出版社,2006,3-138.
15.刘志强.贝叶斯网络与认知图[J],自动化学报,2001,27(4)：52-55.
16. Dawid A P. Statistical theory,the prequential approach[J],Journal of the Royal Statistical Society,1984,147(2):278-292.
17.刘启元,张聪等.信度网近似推理算法(上)[J],计算机科学,2001,28(1)：70-73.
18.刘伟娜,霍利民,张立国.贝叶斯网络精确推理算法的研究[J],微计算机信息,2006,22(3)：92-94.
19.阎平凡.不精确推理中的信度网络方法[J],计算机科学,1989,5,53-56.
20.刘振,史建国,代进进.引入证据后的贝叶斯网络信息传递方法[J],计算机应用,2007,27：85-86.
21.俞奎.贝叶斯网络建模及推理算法研究[D],合肥：合肥工业大学,2007.
22.董立岩,苑森淼,刘光远.基于预测能力的连续贝叶斯网络结构学习[J],计算机工程与应用,2007,43(9)：23-25.
23.Chickering D M.Learning Bayesian Networks is NP-complete[J],learning from data:AI and Statistics,New York:Springer,1996:121-130.
24. Heckerman D.A tutorial on learning with Bayesian network (Technical Report MSR TR-95-06)[M],Redmond:Microsoft Research,1995,42-49.
25.Cheng J.Learning Bayesian networks from date[D],An information theory based approach,Unirersity of lster U K,1998:53-67.
26.王双成,林士敏等.贝叶斯网络结构学习分析[J],计算机科学,2002,27(10)：77-79.
27.王双成,林士敏等.贝叶斯学习、贝叶斯网络与数据采掘[J],计算机科学,2000,27(10)：69-72.
28.王玮,蔡莲红.贝叶斯网络拓扑结构确定方法的研究[J],小型微型计算机系统,2002,23(4)：435-437.
29.林士敏,王双成,陆玉昌.Bayesian方法的计算学习机制和问题求解[J],清华大学学报(自然科学版),2000,40(9)：61-64.
30.余志生.汽车理论[M],北京：机械工业出版社,2000,10-89.
31.齐晓杰.制动系统[M],北京：化学工业出版社,2005,4-108.
32.Shenoy P P,Shafer G.Axioms for probability and belief function propagation[J], Uncertainty in artificial intelligence,Amsterdam:Elsevier Science Publishers,1990, 169-198.
33.J,Pearl.Probabilistic reasoning in intelligent system:Networks of plausible inference[M], San Mateo,CA:Morgan Kaufmann Publishers,1988,44-48.
34. Xiao-dan Zhang,Hai Zhao.A General Situation Assessment Model Based on Rule Extraction[J],Proceedings of the 7th International Conference on Information Fusion, Stockholm,Sweden,Jun.2004,pp:1038-1043.
35.王清印,崔援民,赵秀恒等.预测与决策的不确定性数学模型[M],北京：冶金工业出版社,2001.
36.王道平,冯振声等.故障诊断专家系统中不确定性问题[J],计算机工程与应用,1998,34(12)：33-35.
37.李海成.一种基于规则库的贝叶斯网络开发器的设计与实现[J],陕西科技大学学报,2005,23：93-96.
38.吴泰平.基于故障树分析法与模糊理论的汽车制动系统故障诊断研究[D],广西：广西大学,2004.
39.彭亚敏.模糊逻辑在汽车故障诊断技术中的应用[J],计算机与数字工程,2002,10(4)：24-29.
40.刘喜红,吴康雄.模糊逻辑系统在机械故障诊断中的应用[J],设计与研究,2006,33(2)：15-17.
41.操召发.基于模糊逻辑推理故障智能诊断系统及其仿真研究[D],上海：上海海事大学,2004.