基于IMF熵价值的轮对轴承故障自适应诊断

Adaptive Diagnosis of Wheel-set Bearing Faults Based on IMF's Entropy Value

针对集成经验模态分解(EEMD)方法中本征模态函数(IMF)不能自主筛选的问题,提出IMF价值评价方法,以此评价IMF价值高低。将IMF能量熵作为IMF价值高低的核心评价标准,并基于此建立轮对轴承故障自适应诊断模型。该模型将轴承振动信号进行EEMD分解得到不同尺度的IMF,依据IMF熵价值算法,筛选出价值更高的IMF进行信号重构,对重构信号进行希尔伯特变换,应用其边际谱提取轮对轴承振动特性频率。应用无故障轴承及三种不同故障轴承对本模型进行试验验证。结果表明,该方法能凸显轴承特性频率,能够有效提取轴承旋转频率倍频、故障特征频率及其倍频,并且轴承垂向和横向振动对轴承故障特征频率的检测在谱分辨率及故障表征上都有较好的表现力。

In view of the problem that the intrinsic mode function（IMF） of the integrated empirical mode decom- position（EEMD） method cannot be independently selected, an IMF value evaluation method was put forward to evaluate the value of IMF in this paper. The energy entropies of IMF were taken as the core evaluation crite-ria to measure the use value of IMF and an adaptive fault diagnosis model of wheelset bearing was established based on this conclusion. Under this model, the vibration signals of bearings were decomposed by EEMD to obtain the IMF of different scales. Based on the IMF entropy value algorithm, the higher value IMF was screened to reconstruct the signal, and the reconstructed signal was transformed by Hilbert transforming. The marginal spectrum of the reconstructed signal through Hilbert transforming was applied to extract the vibration characteristic frequency of the wheelset. The model was verified by the test on fault-free bearings and tests on bearings with three diferent faults. The results show that the method can highlight the characteristic frequency of the bearings, and can effectively extract the fundamental and multiplier frequency of rotation and fault features. The method shows good performance in spectral resolution and failure signature in the fault diagnosis of the vertical and horizontal vibration of wheelset bearing.