（南京航空航天大學機械結構力學及控制國家重點實驗室，210016）

摘要：提出了一種基于信號包絡的退化特征量提取方法，對滾動軸承全壽命周期振動信號進行EMD分解，找出各階段與原始信號相關度最大的內蘊模式分量，對此內蘊模式分量進行包絡譜分析，預測滾動軸承故障開始發生的時間及部位，并計算此內蘊模式分量包絡的幅值均值，將其作為刻畫軸承健康狀態的退化特征量，形成退化特征序列，根據經驗設定軸承失效對應的退化特征量閥值。用退化特征量序列訓練新陳代謝灰色模型，用此模型預測退化特征量的變化趨勢，估計退化特征量到達閥值的時間，并據此來預測滾動軸承的疲勞壽命。通過對ZA-2115雙列軸承試驗分析，結果表明，此種退化特征量提取方法結合灰色預測方法可以有效地預測出滾動軸承故障開始發生的時間、部位以及疲勞壽命。
關鍵詞：內蘊模式分量，相關度，包絡分析，灰色模型，故障預測
Abstract: A method of extracting regression features based on the signal envelope is proposed. The vibration signals of the bearing in the whole life cycle are decomposed by the EMD, then find the Intrinsic Mode Function(IMF) which is most corresponding to the vibration signals, then analysis the envelope spectrum of the Intrinsic Mode Function to predict the time when the damage occurs and the location of the damage, and calculate the mean value of the IMF’s envelope amplitude, which is used as the regression feature reflecting the bearing health condition., and then the regression feature list is obtained. A threshold value of the regression feature corresponding with the fatigue failure of the bearing is determined empirically. Then the list is used to train the Grey Model. The regression feature trend is estimated by the trained Grey Model. The time of the regression feature reaching the threshold is estimated and will be used to evaluate the fatigue life of the bearing. The prediction method is validated by vibration signals of ZA-2115 bearing, it shows that the proposed method can provide a proper result for predicting the time of the damage occurring ,the location of the damage, and the fatigue life of the bearing. 
Key words: intrinsic mode function(IMF); degree of correlation; envelope analysis; grey model(GM); damage prognosis

參考文獻
[1] Vachtsevanos G, Lew is F, Roemer M, et al. Intelligent Fault Diagnosis and Prognosis for Engineering Systems[M]. Hoboken, NJ:Wile,2006.
[2] C. James Li, Hyungdae Lee. Gear fatigue crack prognosis using embedded model, gear dynamic model and fracture mechanics[J]. Mechanical Systems and Signal Processing 19(2005) 836-846.
[3] Xuefei Guan, Ratneshwar Jha, Yongming Liu. Probabilistic fatigue damage prognosis using maximum entropy approach[J]. J Intell Manuf (2012) 23:163-171.
[4] Achmad Widodo, Bo-Suk Yang. Machine health prognostics using survival probability and support vector machine[J]. Expert Systems with Applications 38(2011) 8430-8437.
[5] Wahyu Caesarendra, Achmad Widodo, Bo-Suk Yang. Combination of probability approach and support vector machine towards machine health prognostics[J]. Probabilistic Engineering Mechanics 26(2011) 165-173.
[6] Nagi Gebraeel, Mark Lawley, et al. Residual life predictions from vibration-based degradation signals: A neural network approach[J]. IEEE Transactions on Industrial Electronics, 2004, 51(3): 694-700.
[7] 王紅軍，張建民，徐小力. 基于支持向量機的機械系統狀態組合預測模型研究[J]. 振動工程學報，2006,19(2): 242~245.
[8] 曾慶虎，邱靜，劉冠軍. 基于小波相關特征尺度熵的HSMM設備退化狀態識別與故障預測方法研究[J].儀器儀表學報，2008, 29(12): 2559~2564.
[9] Lee, J., Qiu, H., Yu, G., Lin, J.: Rexnord Technical Services: Bearing Data Set, IMS, Univ. of Cincinnati. NASA Ames Prognostics Data Repository (2007), http://ti.arc.nasa.gav/project/prognostic-data-repository.
[10] Ruoyu Li, Ponrit Sopon, David He. Fault features extraction for bearing prognostics[J]. J Intell Manuf, DOI 10.1007/s10845-009-0353-z.
[11] 劉思峰，謝乃明等. 灰色系統理論及其應用(第四版)[M]. 科學出版社，2008.

第一作者簡介：
張登，男，1988年生，南京航空航天大學機械結構力學及控制國家重點實驗室，振動工程研究所，碩士生，