作为表征信号冲击特征的常用指标，峭度容易受到少量大幅值异常数据或随机冲击的干扰，非常不利于滚动轴承、齿轮等旋转机械故障特征的提取。为更有效地表征信号的循环冲击特征，将线性峭度概念引入到滚动轴承故障特征表征中。同时，结合经验小波变换提出线性峭度图方法。该方法通过构造具有紧支撑的经验小波函数，将信号进行正交分解；随后，计算不同分解方案下子信号的线性峭度构建线性峭度图；最终，选择线性峭度最大的子信号进行平方包络分析实现故障特征的提取与增强。通过滚动轴承内外圈剥落故障的仿真与试验分析表明：线性峭度比峭度更适合表征信号的循环冲击特征。与基于传统峭度图方法和子频带谱峭度平均的峭度图方法相比，提出的线性峭度图方法在信号的分解与表征两方面都进行了改进，具有更强的循环冲击特征提取能力，具有更高的工程应用价值。但线性峭度图方法在滤波器组构建上效率相对较低，还有很大的提升空间。
Abstract
As one of the common used statistic indexes for impulse characterizing, kurtosis is inclined to be disturbed by the outlines or stochastic impulse with large value and is not convenient for the feature extraction of the rotating machinery, such as rolling element bearings and gears.To accurately characterizing the feature of signals with cyclic impulses, the L-kurtosis was introduced to the fault diagnosis of rolling element bearings and a novel feature extraction method, named as L-kurtogram, was proposed on the basis of empirical wavelet transform (EWT).By constructing a tight supported empirical wavelet, the original signal can be orthogonally decomposed to be several sub-signals.Then, the L-kurtogram was constructed by calculating the L-kurtosis of the sub-signals of different decompositions.Finally, the cyclic impulse feature was extracted by calculating squared envelope spectrum of the sub-signal with maximum L-kurtosis.Both simulations and experiments were used to validate the efficacy of the proposed method.It is shown that the L-kurtosis is more suitable than the kurtosis for the impulse feature characterization of the signal with cyclic impulses.Furthermore, compared with the original kurtogram and the fast kurtogram based on sub-frequency-band spectral kurtosis average, the L-kurgram is more powerful for the cyclic impulse feature extraction and is more valuable for the engineering application, since the improvements in both signal decomposition and the impulse feature characterization.However, the consuming time for the construction of the filter bank of L-kurtogram is much longer than that of the fast kurtogram and is the major investigation context in the future.
关键词
线性峭度 /
线性峭度图 /
冲击特征 /
特征提取 /
滚动轴承
{{custom_keyword}} /
Key words
L-kurtosis /
L-kurtogram /
impulse feature /
feature extraction /
rolling element bearings
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] Dyer D, Stewart R M. Detection of rolling element bearing damage by statistical vibration analysis [J]. Journal of mechanical design, 1978, 100(2): 229-235.
[2] Tandon N. A comparison of some vibration parameters for the condition monitoring of rolling element bearings [J]. Measurement, 1994, 12(3): 285-289.
[3] Vass J, Šmíd R, Randall R B, et al. Avoidance of speckle noise in laser vibrometry by the use of kurtosis ratio: Application to mechanical fault diagnostics [J]. Mechanical Systems and Signal Processing, 2008, 22(3): 647-671.
[4] Liu B. Selection of wavelet packet basis for rotating machinery fault diagnosis [J]. Journal of Sound and Vibration, 2005, 284(3-5): 567-582.
[5] Zhang D, Sui W. The optimal morlet wavelet and its application on mechanical fault detection[C]//Wireless Communications, Networking and Mobile Computing, 2009. WiCom'09. 5th International Conference on. IEEE, 2009: 1-4.
[6] Hong H, Liang M. K-hybrid: a kurtosis-based hybrid thresholding method for mechanical signal denoising [J]. Journal of vibration and acoustics, 2007, 129(4): 458-470.
[7] Antoni J. The spectral kurtosis: a useful tool for characterising non-stationary signals [J]. Mechanical Systems and Signal Processing, 2006, 20(2): 282-307.
[8] Bozchalooi I S, Liang M. A smoothness index-guided approach to wavelet parameter selection in signal de-noising and fault detection [J]. Journal of Sound and Vibration, 2007, 308(1-2): 246-267.
[9] McDonald G L, Zhao Q, Zuo M J. Maximum correlated Kurtosis deconvolution and application on gear tooth chip fault detection [J]. Mechanical Systems and Signal Processing, 2012, 33: 237-255.
[10] 代士超，郭瑜，伍星，那靖. 基于子频带谱峭度平均的快速谱峭度图算法改进. 振动与冲击, 2015, 34(7): 98-102.
DAI Shichao ，GUO Yu， WU Xing，et al. Improvement on fast kurtogram algorithm based on sub-frequency-band spectral kurtosis average [J]. Journal of Vibration and Shock，2015, 34(7): 98-102.
[11] Ulrych T J, Velis D R, Woodbury A D, Sacchi M D. L-comments and C-comments. Stochastic Environmental Research and Risk Assessment, 2000, 14: 50-68.
[12] Hosking J R M. L-moments: Analysis and Estimation of Distributions using Linear Combinations of Order Statistics, Journal of the Royal Statistical Society，1990, 52(1):105-124.
[13] Greenwood J A, Landwehr J M, Matalas N C, et al. Probability weighted moments: definition and relation to parameters of several distributions expressable in inverse form [J]. Water Resources Research, 1979, 15(5): 1049-1054.
[14] Wang Q J. Direct sample estimators of L moments [J]. Water resources research, 1996, 32(12): 3617-3619.
[15] Gilles J. Empirical wavelet transform [J]. IEEE transactions on signal processing, 2013, 61(16): 3999-4010.
[16] Liu S., Hou S., He K., Yang W., L-Kurtosis and its application for fault detection of rolling element bearings, Measurement (2017), doi: https: // doi.org/10.1016/j.measurement.2017.11.049
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(2464 KB)
