Study on Feature Extraction and Diagnosis Method of Rolling Bearing Faults Based on EMD and KPCA
-
摘要: 针对滚动轴承故障种类识别与程度判断问题,提出了一种融合经验模式分解与核主元分析的故障诊断方法:首先,运用经验模式分解将滚动轴承故障信号分解成不同特征尺度下的本征模式分量,采用Hilbert-Huang变换对信号进行相应的时频分析,从本征模式分量函数和瞬时频率中分别提取时域和频域的统计特征集与无量纲特征集;其次,引入非线性核主元分析方法,对故障特征集进行处理,从而消除特征集中的冗余特征,并大幅度降低特征向量维数,得到能够反映故障本质的主元特征集;最后,构造支持向量机多类分类网络,实现了对不同故障模式与不同损伤程度滚动轴承的故障诊断。Abstract: A new diagnosis method for the fault classification and damage degree of rolling bearing based on empiricalmode decomposition(EMD) and kernel principal component analysis(KPCA) is proposed.Firstly, the fault signal ofrolling bearing was decomposed into several intrinsic mode functions under different characteristic scales with EMD.ThenHilbert-Huang transform was employed to calculate the statistical features and dimensionless features in both time domainand frequency domain.Secondly, KPCA based on samples was introduced to eliminate redundant features and reduce thevector dimension greatly.Thus, the KPCA feature vector was acquired, which could reflect the failure.Furthermore, theclassifier based on one-versus-one support vector machine was constructed.At last, the classification of three failuremodes and different damage degree for rolling bearing was completed.
-
[1] Wen C T,Li Y F,Duc D L,et al.An insight conceptto select appropriate IMFs for envelope analysis ofbearing fault diagnosis[J].Measurement,2012,45(6):1489-1498 [2] 张明,李志勇,崔帅,等.航空航天轴承接触疲劳寿命分析[J].机械科学与技术,2012,31(6): 938-941Zhang M,Li Z Y,Cui S,et al.Analysis on contactfatigue life of aerospace bearing[J].Mechanical Scienceand Technology,2012,31(6): 938-941 (in Chinese) [3] 李琳,张永祥,明廷涛.EMD 降噪的关联维数在齿轮故障诊断中的应用研究[J].振动与冲击,2009,28(4): 145-148Li L,Zhang Y X,Ming Y T.Gear fault diagnosis basedon correlation dimension pre-processed with EMD[J].Journal of Vibration and Shock,2009,28 (4): 145-148(in Chinese) [4] 王晓冬,何正嘉,訾艳阳.多小波自适应构造方法及滚动轴承复合故障诊断研究[J].振动工程学报,2010,23(4): 438-444Wang X D,He Z J,Zi Y Y.Adaptive construction ofmulti-wavelet and research on composite fault diagnosisof rolling bearing[J].Journal of Vibration Engineering,2010,23(4): 438-444 (in Chinese) [5] 雷亚国,何正嘉.混合智能故障诊断与预示技术的应用进展[J].振动与冲击,2011,30(9): 129-134Lei Y G,He Z J.Advances in applications of hybridintelligent fault diagnosis and prognosis technique[J].Journal of Vibration and Shock,2011,30 (9): 129-134(in Chinese) [6] Huang N E,Shen Z,Long S R,et al.The empiricalmodel decomposition and hilbert spectrum for nonlinearand non-stationary time series analysis [C]//Proceedings of the Royal Society Lond,1998,454:903-995 [7] Peng Z K,Tse P W,Chu F L.A comparison study ofimproved Hilbert-huang transform and wavelettransform: application to fault diagnosis for rollingbearing[J].Mechanical Systems and Signal Processing,2005,19(5): 974-988 [8] 张德祥,汪平,吴小培,等.基于EMD 和非线性峭度的齿轮故障诊断[J].振动、测试与诊断,2012,32(1):56-61Zhang D X,Wang P,Wu X P,et al.A diagnosis ofbear fault based on EMD and nonlinear kurtosis[J].Journal of Vibration Measurement & Diagnosis,2012,32(1): 56-61 (in Chinese) [9] 刘然,许宝杰.基于EMD 和全息谱的设备故障诊断方法研究[J].机械科学与技术,2011,30 (11):1922-1926.Liu R,Xu B J.The study of the equipment failurediagnosis method based on EMD and holospectrum[J].Mechanical Science and Technology,2011,30 (11):1922-1926 (in Chinese) [10] 雷亚国,何正嘉,訾艳阳.基于混合智能新模型的故障诊断[J].机械工程学报,2008,44(7): 112-117Lei Y G,He Z J,Zi Y Y.Fault diagnosis based onnovel hybrid intelligent model[J].Chinese Journal ofMechanical Engineering,2008,44 (7): 112-117 (in Chinesee) [11] Lei Y G,He Z J,Zi Y Y,et al.Fault diagnosis ofrotating machinery based on multiple ANFIScombination with GAs[J].Mechanical Systems andSignal Processing,2007,21(5): 2280-2294 [12] Chakraborty S,Yeh C H.A simulation comparison ofnormalization procedures for topsis[C]//InternationalConference on Computers and IndustrialEngineering,2009 [13] 蒋维杨,赵嵩正,刘丹,等.大样本评价的定量指标无量纲化方法[J].统计与决策,2012,17: 5-11Jiang W Y,Zhao S Z,Liu D,et al.A method ofquantitative undimensionalization in the evaluation of alarge sample[J].Statistics and Decision,2012,17: 5-11(in Chinese) [14] 赵立杰,柴天佑,王纲.多元统计性能监视和故障诊断技术研究发展[J].信息与控制,2004,33(2): 197-200Zhao L J,Chai T Y,Wang G.The progress ofmultivariate statistical performance monitoring and faultdiagnosis[J].Information and Control,2004,33 (2):197-200 (in Chinese) [15] Perlibakas V.Distance measures for PCA-basedrecognition[J].Pattern Recognition Letters,2004,25: 711-724 [16] Detroja K P,Gudi R D,Patwardhan S C.Data reductionalgorithm based on principle of distributional equivalencefor fault diagnosis[J].Control Engineering Practice,2012,20(10): 1033-1041 [17] Schlkopf B,Sola A,Muller K R.Nonlinear componentanalysis as a kernel eigenvalue problem [J].NeuralComputation,1998,10(5): 1299-1319 [18] Vapnik V,Golowich S,Alex S,et al.Support vectormethod for function approximation regression estimationand signal processing [J].Advances in NeuralInformation Processing Systems,1996,(9): 281-287 [19] Loparo K A.Bearings vibration data set,case westernreserve university[DB/OL].http: //www.eecs.cwru.edu /laboratory /Bearing /download.htm1524
点击查看大图
计量
- 文章访问数: 107
- HTML全文浏览量: 29
- PDF下载量: 4
- 被引次数: 0