基于SVM的航空发动机油样光谱诊断 界限值制定

李爱; 陈果

基于SVM的航空发动机油样光谱诊断界限值制定

李爱,
陈果

南京航空航天大学民航学院,南京 210016

基金项目: 成都飞机工业(集团)有限责任公司项目

计量
- 文章访问数: 1560
- HTML浏览量: 29
- PDF量: 21
- 被引次数: 0
出版历程
- 收稿日期: 2010-03-17
- 修回日期: 2010-07-01
- 刊出日期: 2011-04-28

Establishment of the threshold of oil spectrum analysis in the aircraft engine based on SVM

LI Ai,
CHEN Guo

College of Civil Aviation, Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China

摘要

摘要: 运用支持向量机(SVM)估计航空发动机油样光谱数据的概率密度函数,根据光谱数据的概率分布求出航空发动机光谱诊断各金属元素的质量分数、质量分数梯度及质量分数比例的正常、警告、以及异常界限值.利用实际的航空发动机光谱数据进行了研究,并与传统的基于正态分布假设下所确定的界限值进行了比较分析.结果表明,实际的航空发动机光谱数据并不完全服从正态分布假设,因此基于SVM的油样光谱界限值制定方法所得到的结果更加真实可靠.
- 航空发动机 /
- 光谱诊断 /
- 支持向量机 /
- 概率密度估计 /
- 界限值
Abstract: The probability density of the aircraft engine oil sample spectral data was estimated by using support vector machines(SVM).According to the probability distribution of spectral data,the normal,warning and abnormal threshold of the mass fraction,mass fraction gradient and the mass fraction ratio of aircraft engine were calculated based on spectroscopic diagnosis.This method was also verified by using the actual spectral data in aircraft engine.Through comparison and analysis of the threshold based on the normal distribution assumption,it shows that the actual spectral elements of aircraft engine are not entirely subject to normal distribution assumption,so the results established by SVM-based oil-like spectral threshold method are more genuine and reliable.
- aircraft engine /
- spectral diagnostics /
- support vector machine(SVM) /

HTML

参考文献(12)

[1]	任国全,张培林,张英堂.装备油液智能监控原理[M].北京:国防工业出版社,2006.
[2]	李柱国.机械润滑与诊断 [M].北京:化学工业出版社, 2005.
[3]	Duda,Hart,Stork.模式分类[M].李宏东,姚天翔,译.北京:机械工业出版社,2006.
[4]	Miller G,Horn D.Probability density estimation using entropy maximization[J].Neural Computation,1998,10(7),1925-1938.
[5]	Vapnik.统计学习理论[M].徐建华,张学工,译.北京:电子工业出版社,2009.
[6]	Vapnik V,Golowich S E,Smola A.Support vector method for function approximation,regression estimation,and signal processing[J].Advances in Neural Information Processing Systems,1996,9:281-287.
[7]	王文清.机械装备光谱油样分析故障诊断中界限值制定与知识库建立的研究与应用 .北京:北京理工大学, 1994. WANG Wenqing.The threshold formulation and knowledge base in fault diagnosis by oil spectrometer .Beijing:Beijing Institute of Technology,1994.(in Chinese)
[8]	Weston J,Gammerman A,Stitson M O,et al. Support vector density estimation[M]. Cambridge,MA:MIT Press, 1999.
[9]	Vapnik V N,Mukherjee S.Support vector method for multivariate density estimation[J].Advances in Neural Information Processing Systems,2000,12:659-665.
[10]	张炤.基于支持向量机的概率密度估计方法[J].系统仿真学报,2005,17(10):2355-2357. ZHANG Zhao.Density estimation based on support vector machines[J].Journal of System Simulation,2005,17(10):2355-2357.(in Chinese)
[11]	Cristranini,Shawe-Yaylor.支持向量机导论[M].李国正,王猛,曾华军,译.北京:电子工业出版社,2004.
[12]	Parzen E.On estimation of a probability density function and mode[J].Annals of Mathematical Statistics,1962,33(3):1065-1076.