摘要
航空发动机是民航飞机的动力核心,因此对其运行状态进行监测至关重要。在磨损故障监测中,针对不同的监测对象,有选择地联合运用几种油液监测手段,是获得更为准确分析结论的有效途径。本文以XRF技术和显微图像技术为基础,引入神经网络、灰色系统理论、概率统计、D-S证据理论等智能方法,对航空发动机的智能监测技术进行研究,致力于研制出一套智能化、系统化的智能监测系统。
XRF(X射线荧光)技术利用同位素放射源辐射被测物质,通过一系列装置产生能谱,根据能谱来判别物质的元素类别。显微图像技术通过显微光学成像系统、图像采集与分析处理等过程,实现对发动机磨损磨粒的监测,具有颗粒计数器与铁谱仪的双重功能。
本文研究工作主要包括以下三部分:首先,在基于XRF分析技术得到能谱数据的基础上,对能谱进行处理和分析,得到元素的浓度;研究了金属元素浓度的界限值的制定方法,以及在此基础上的故障诊断,并运用灰色预测实现了发动机的故障预测。其次,在显微图像分析得到磨损磨粒参数以及磨粒数目的基础上,实现了运用径向基函数神经网络(RBF)、BP神经网络和灰度关联分析对磨损磨粒的融合识别,提高了磨粒识别的准确率。在磨粒识别的基础上实现了对发动机故障进行监测。最后,提出了基于以上两种分析的融合诊断方法以及智能综合分析模型,并开发了智能监测系统软件。
Aeroengine supplies the major motive power to the modern aircraft, so the status inspection of the aero-engine operation plays undoubtedly an important role. According to different objects, the conclusion of fuseing several different monitoring means is more accurately. This paper beased on XRF (X-ray fluorescence) technology and introducing artificial neural network,gray system theory,probability statistics,Dempster-Shafer evidential theory and other intelligent calculation,in order to develop a intelligent monitoring system.
XRF (X-ray fluorescence) technology, using isotope radioactive sources to eradiate materials which to be measured, produces energy spectrum through a series of installations, which is used to determine the elemental material categories. Microscopic image technique which with the functions of particle counter and ferrograph , achieved the diagnosis on the aero-engine wear faults by useing mcroscopic optical imaging system and digital image processing method.
The research work mainly includes the following three aspects. Firstly, processed XRF energy spectrum and out putted the element concentrations. Researched the methord of calculating the threshold value of the concentration of the wear debris and forecasted the wear fault by grey prediction. Secondly, based on the number and parameters of wear debris, fusing the radial based function (RBF) neural network, back-propagation (BP) neural network and grey relational analysis methods, improving the accuracy rate of debris auto-recognition based on the collected information of the debris configuration characteristics; and conducting the diagnosis on the aero-engine wear faults according to the recognition.Finally, a synthetic intelligent diagnosis model by data fusion fault diagnosis method based on above diagnosis method was put forward, an intelligent monitoring software system of aero-engine is developed based on the above research.
引文
[1]张宝珍,世界民航维护、修理和大修业发展趋势.数控机床市场,2006(10):30～32.
[2]方昌德.发动机状态监测和故障诊断系统的发展.国际航空,2005,6.
[3]范作民,孙春林,白杰著.航空发动机故障诊断导论.北京:科学出版社,2004.
[4]王国庆,润滑油液监测技术现状与发展.润滑油,2004,19(5):7～11.
[5]董宏军,赵海波等.我国油液分析技术现状及发展趋势.林业机械与木工设备,2004,32 (6):10～11.
[6]姜旭峰,费逸伟等.滑油监测技术在航空发动机预防性维修中的应用.润滑与密封,2004,(2):56～58.
[7]严新平.从Technology Showcase_2000国际会议谈油液监测技术的发展.状态监测与诊断技术,2000.8.
[8]黎琼炜,新型油液在线监控技术,测控技术,2005,24(4):6～10.
[9] Gary R.Humphrey,RobertWhitlock,etal.Energy Dispersive X-Ray Fluorescence Evaluation of Debris from F-18 Engine Oil Filters,1998.
[10] A.Zararsiz,R.Kirmaz,etal.Determination of wear metals in used lubricating oils by X-ray fluorescence spectrometry,Nuclear Instruments and Methods in Physics Research,1996.
[11]王坚,张英堂.油液分析技术及其在状态监测中的应用.润滑与密封,2002 (4):77～78.
[12]张晨辉,林亮智编著.设备润滑与润滑油应用.北京:机械工业出版社,1994.
[13]王群伟,顾大强.机械设备磨损状态的远程监测诊断系统的研究.润滑与密封,2002(5):49～52.
[14]张焕梅.大型机械设备润滑油样分析.露天采煤技术,1997(1):27～29.
[15]袁梅,吕俊芳等.IXRF法用于航空发动机磨损在线监测的研究.北京航空航天大学学报,2000,26(2):153～155.
[16]张家骅,徐君权等编著.放射性同位素X射线荧光分析.北京:原子能出版社,1981.
[17] Gary R.Humphrey,Robert Whitlock,Darrell B.Churchill.The Path to Affordable Long Term Failure Warning:The XRF-Wear Monitor,1998,6.
[18]袁梅,张俊芳等.航空发动机磨损在线监测的能谱数据处理方法研究.仪器仪表,2000, 21(2):173～176.
[19]曾国强,林延畅等.可视化XRF软件数据采集系统的开发.物探化探计算技术,2004, 26(3):274～277.
[20]郭余峰,自然伽马能谱的平滑滤波处理.大庆石油学院学报,2003, 27(3):113～117.
[21]段再煜,陈建华等.基于Matlab平台上γ能谱光滑处理.核动力工程,2006, 28(3):125～127.
[22]尚凤军,王海霞,周蓉生.X荧光谱分析中寻峰算法的探讨及实践.物探化探计算技术2000,22(4).
[23] Dennis J. Kalnicky,Raj Singhvi. Field portable XRF analysis of environmental samples. Journal of Hazardous Materials,2001,93～122.
[24]吴振锋,左洪福,韩丽枫.磨损颗粒形态学智能分析技术.南京航空航天大学学报, 1999年增刊: 7～12
[25]葛世荣,索双富.表面轮廓分形维数计算方法的研究.摩擦学学报,1997,4(17):354～362.
[26]吴振锋.基于磨粒分析与信息融合的发动机磨损故障诊断技术研究[D].南京航空航天大学.
[27]汤兵兵,张来斌,樊建春.计算机磨粒识别技术研究进展[J].润滑与密封,2001,26(6): 36～37.
[28] Ge Shirong,Chen Guoan,ZhangXiaoyun.Fractal characterlzation of Wear particle accumulation in the wear Process.Wear.2001,251:1227～1233.
[29] C.Q.Yuan,J.Li,X.P.Yan,Z.peng. The use of the fractal description to Characterize engineering surfaces and wear Particles. Wear. 2003,255:315～326.
[30]王伟华,殷勇辉,王成焘.基于径向基函数神经网络的磨粒识别系统[J].摩擦学学报, 2003,23(4): 340～343.
[31]康剑莉,芦亚萍,周银生.改进的BP算法在磨粒识别中的应用[J].润滑与密封, 2005 (3):149～150.
[32]陈果,左洪福.润滑油金属磨粒的分类参数研究[J].航空学报,2002,23 (3):279～281.
[33]杨其明.磨粒分析[M].北京:中国铁道出版社,2002:1～50.
[34] Kolmogorov. On the representation of continuous functions of many variables by superposition of continuous functions of one variable and condition [J]. Dok1, Akad. Nauk,USSR,1957,114: 953～956.
[35]邓聚龙.灰色系统理论教程[M].武汉:武汉理工大学出版社, 1990,33～851.
[36]廖振强,张培林,汪传忠.铁谱磨粒形态分形特征参数提取方法研究.润滑与密封, 2004(6)
[37]姜旭峰,彭秀华,费逸伟,李华强.基于灰色关联分析的磨粒识别[J].润滑与密封,2005,(5):112～116.
[38]康耀红著.数据融合理论与应用.西安:西安电子科技大学出版社,2006.
[39] P.Podsiadlo, G.W.Staehowiak.Scale-invariant analysis of wear Partiele surface morphology I:Theoretieal baekground,computer implementation and technique testing.Wear,2000,242:160～179.
[40] G.W.Stachowiak,P.Podsiadlo.Surface charaeterization of wear Particles.Wear,1999,225-229:1171～1185.
[41] G.B.Staehowiak,G.W.staehowiak. The effects of partiele characteristics on three-body abrasive wear,Wear,2001,249:201～207.
[42] Z.Peng,T.B.Kirk. Wear Particle classification in a fuzzy grey system,Wear,1999,225-229:1238～1247.
[43] P.Podisiadlo,G.W.Staehowiak. Scale-invariant analysis of wear partiele surface morphology III:Pattern recognition,Wear,2000,242:189～201.
[44]虞和济,陈长征,张省等著.基于神经网络的智能诊断[M].北京:电子工业出版社,2007.
[45]杜文吉,谢维信.D-S证据理论中的证据融合.系统工程与电子技术, 1999.12(21): 92～94.
[46]赵新泽.油液监测中信息融合的思想与方法研究.武汉水利电力大学学报, 2000,2(22): 156～159.
[47]孙全,叶秀清,顾伟康.一种新的基于证据理论的合成公式[J].电子学报,2000, 28 (8) : 117～119.
[48]李弼程,王波,魏俊.一种有效的证据理论合成公式[J].数据采集与处理,2002,17 (1): 33～36.
[49] Yager R R. On the Dempster-Shafter framework and new combination rules [J]. Information Sciences, 1987, (41):93～137.
[50] Kai G. Architecture and design of a diagnostic information fusion system[J]. Artificial Intelligence for Engineering Design, Analysis and Manufacturing, 2001, (15):335～348.
[51] Yager R R. Fuzzy modeling for intelligent decision making under uncertainty [J]. IEEE Tran s on SMC, Part B, 2000, 30(1):60～70.
[52]树悦,陈东林.航空发动机磨损趋势预测的GM (1,1)模型方法.润滑与密封,2005(5):96～100.
[53]罗运柏,于萍等.用灰色模型预测变压器油中溶解气体的含量.中国电机工程学报,2001, 21(3):65～69.
[54]郑长松,马彪.基于改进欧拉算法的灰预测模型在综合传动故障预测中的应用研究.润滑与密封,2007, 32(5):45～47.
[55]万耀青,郑长松等.原子发射光谱仪作油液分析故障诊断的界限值问题.机械强度,2006,485～487.
[56]左洪福编著.发动机磨损状态监测与故障诊断技术.北京:航空工业出版社, 1995.
[57]陈立波,张占刚.应用滑油光谱监控技术成功预报的一起航空发动机异常磨损故障.材料工程,2003增刊:146～147.
[58]姚红宇等.民航飞机失效分析实例选编.北京:中国民航出版社,2003
[59] Chen Guo,Yang Yuwei,etal. Intelligent Fusion for Aeroengine Wear Fault Diagnosis, Transaction of Nanjing University of Aeronautics & Astronautics, Dec.2006.
[60]萨师煊,王珊.数据库系统概论.北京:高等教育出版社,2000.
[61]刘斌,王忠.面向对象程序设计—Visual C++[M].北京:清华大学出版社,2003
[62]宋好好,张明强等.对用Chart控件的SetData函数绘制曲线图和散点图的研究与实现.山东电子,2003(2).
[63] Steven Bogaerts,David Leake.A Framework For Rapid And Modular Case-Based Reasoning System Development Report Version 1.0*,Bloomington,Indiana,USA: Indiana University,2005.
[64]成功,杨佃福.VC中几种数据库访问技术的比较与选择.计算机应用与研究,2002, 19(2):82～84.
[65] Jon Bates,Tim Tompkins,何健辉等译.实用Visual C++6.0教程.北京:清华大学出版社,2000.
