基于EMD的灰色模型的疲劳剩余寿命预测方法研究
|
摘要
工程上的振动信号多为非线性非平稳信号,为了利用工程振动信号预测机械产品的疲劳剩余寿命,提出改进的经验模态分解方法对振动信号进行分解,分离故障特征频率到某本征模态函数中,计算全寿命周期各阶段故障特征频率所在本征模态函数的均方根值、峭度等时域特征指标,将其作为刻画机械产品健康状态的退化特征量,形成退化特征量序列,根据经验设定机械产品完全失效对应的退化特征量阈值。用退化特征量序列训练灰色模型,然后用训练好的灰色模型预测退化特征量的变化趋势,判断不同退化特征量用于刻画机械产品退化过程的可行性,估计可用退化特征量达到退化特征量阈值的时间并据此预测机械产品的剩余疲劳寿命。通过6205深沟球轴承全寿命周期振动信号对其进行验证,结果表明,可用的退化特征量结合该方法可以有效地预测小型球轴承的疲劳剩余寿命。
Most of practical signals are nonlinear and nonstationary.In order to predict the residual fatigue life of equipment based on practical signals,advanced Empirical Mode Decomposition(EMD) is proposed and these practical signals are decomposed by the advanced EMD and then a series of Intrinsic Mode Functions(IMFs) are obtained after the decomposition process.One or more IMFs will include the defect characteristic frequency of the equipment.The characteristic value in time domain of IMFs including defect characteristic frequency such as Root Mean Square or kurtosis and so on is considered as the characteristic quantity of the degradation process of a equipment and the characteristic quantity list of the degradation process will be obtained.Then a threshold value of the characteristic quantity corresponding with the fatigue failure of the equipment is determined empirically.Then the characteristic quantity of IMFs including the defect characteristic frequency in each period of the whole life cycle of the equipment are figured out and a characteristic value list is obtained.Then the list is used to train the Grey Model.The characteristic quantity trend is estimated by the trained Grey Model.The feasibility of degenerative lists based on different characteristic values is estimated.The time of usable degenerative value reaching the threshold value is estimated and will be used to evaluate the remaining fatigue life of the equipment.The prediction method is validated by vibration signals of 6205 deep groove bearing in the whole life cycle,it shows that the prediction method can provide a proper result for predicting the remaining fatigue life of small ball bearings.
Most of practical signals are nonlinear and nonstationary.In order to predict the residual fatigue life of equipment based on practical signals,advanced Empirical Mode Decomposition(EMD) is proposed and these practical signals are decomposed by the advanced EMD and then a series of Intrinsic Mode Functions(IMFs) are obtained after the decomposition process.One or more IMFs will include the defect characteristic frequency of the equipment.The characteristic value in time domain of IMFs including defect characteristic frequency such as Root Mean Square or kurtosis and so on is considered as the characteristic quantity of the degradation process of a equipment and the characteristic quantity list of the degradation process will be obtained.Then a threshold value of the characteristic quantity corresponding with the fatigue failure of the equipment is determined empirically.Then the characteristic quantity of IMFs including the defect characteristic frequency in each period of the whole life cycle of the equipment are figured out and a characteristic value list is obtained.Then the list is used to train the Grey Model.The characteristic quantity trend is estimated by the trained Grey Model.The feasibility of degenerative lists based on different characteristic values is estimated.The time of usable degenerative value reaching the threshold value is estimated and will be used to evaluate the remaining fatigue life of the equipment.The prediction method is validated by vibration signals of 6205 deep groove bearing in the whole life cycle,it shows that the prediction method can provide a proper result for predicting the remaining fatigue life of small ball bearings.
引文
[1]Vachtsevanos G,Lewis F,Roemer M,et al.Intelli-gent Fault Diagnosis and Prognosis for EngineeringSystems[M].Hoboken,NJ:Wile,2006.
[2]Kacprzynski G J,Sarlashkar A,Roemer M J,et al.Predicting remaining life by fusing the physics offailure modeling with diagnostics[J].JOM Journal ofthe Minerals,Metals and Materials Society,2004,56:29—35.
[3]Qiu J,Set B B,Liang S Y,et al.Damage mechanicsapproach for bearing lifetime prognostics[J].Me-chanical Systems and Signal Processing,2002,16:817—829.
[4]Lundberg G,Palmgren A.Dynamic capacity ofrolling bearings[J].Acta Polytechnica MechanicalEngineering Sciences,1947,7(3):1—7.
[5]Howard I.A Review of Rolling Element Bearing Vi-bration Detection,Diagnosis and Prognosis[M].Can-berra(Australia):Defence Science and TechnologyOrganisation,1995.
[6]Norden E H,Zheng S,Steven R L.The empirical de-composition and the Hilbert spectrum for nonlinearand non-stationary time series analysis[A].Proceed-ings of the Royal Society of London[C].London,England,1998:903—995.
[7]Dong X,Yongcheng X,Xun C,et al.Life cycle vi-bration analysis based on EMD of rolling elementbearing under ALT by constant stress[A].Proceed-ings of ICRMS2009[C].Chengdu,China,2009:1 177—1 182.
[8]奚立峰,黄润青,李兴林,等.基于神经网络的球轴承剩余寿命预测[J].机械工程学报,2007,43(10):137—143.
[9]高强,杜小山,范虹,等.滚动轴承故障的EMD诊断方法研究[J].振动工程学报,2007,20(1):15—18.
[10]Shao Y,Nezu K.Prognosis of remaining bearing lifeusing neural networks[J].Journal of Systems andControl Engineering,2000,214(13):217—230.
[11]Gebraeel N,Lawley M,Liu R,et al.Residual lifepredictions from vibration-based degradation signals:A neural network approach[J].IEEE Transactions onIndustrial Electronics,2004,51(3):694—700.
[12]黄大吉,赵进平,苏纪兰.希尔伯特-黄变换的端点延拓[J].海洋学报,2003,25(1):1—10.
[13]龙思胜,张铁宝,龙峰.希尔伯特-黄变换中拟合过冲和端点飞翼的原因及解决办法[J].地震学报,2005,25(9):561—568.
[2]Kacprzynski G J,Sarlashkar A,Roemer M J,et al.Predicting remaining life by fusing the physics offailure modeling with diagnostics[J].JOM Journal ofthe Minerals,Metals and Materials Society,2004,56:29—35.
[3]Qiu J,Set B B,Liang S Y,et al.Damage mechanicsapproach for bearing lifetime prognostics[J].Me-chanical Systems and Signal Processing,2002,16:817—829.
[4]Lundberg G,Palmgren A.Dynamic capacity ofrolling bearings[J].Acta Polytechnica MechanicalEngineering Sciences,1947,7(3):1—7.
[5]Howard I.A Review of Rolling Element Bearing Vi-bration Detection,Diagnosis and Prognosis[M].Can-berra(Australia):Defence Science and TechnologyOrganisation,1995.
[6]Norden E H,Zheng S,Steven R L.The empirical de-composition and the Hilbert spectrum for nonlinearand non-stationary time series analysis[A].Proceed-ings of the Royal Society of London[C].London,England,1998:903—995.
[7]Dong X,Yongcheng X,Xun C,et al.Life cycle vi-bration analysis based on EMD of rolling elementbearing under ALT by constant stress[A].Proceed-ings of ICRMS2009[C].Chengdu,China,2009:1 177—1 182.
[8]奚立峰,黄润青,李兴林,等.基于神经网络的球轴承剩余寿命预测[J].机械工程学报,2007,43(10):137—143.
[9]高强,杜小山,范虹,等.滚动轴承故障的EMD诊断方法研究[J].振动工程学报,2007,20(1):15—18.
[10]Shao Y,Nezu K.Prognosis of remaining bearing lifeusing neural networks[J].Journal of Systems andControl Engineering,2000,214(13):217—230.
[11]Gebraeel N,Lawley M,Liu R,et al.Residual lifepredictions from vibration-based degradation signals:A neural network approach[J].IEEE Transactions onIndustrial Electronics,2004,51(3):694—700.
[12]黄大吉,赵进平,苏纪兰.希尔伯特-黄变换的端点延拓[J].海洋学报,2003,25(1):1—10.
[13]龙思胜,张铁宝,龙峰.希尔伯特-黄变换中拟合过冲和端点飞翼的原因及解决办法[J].地震学报,2005,25(9):561—568.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心