基于形态学滤波和CEEMDAN-WVD的车轮失圆诊断

Wheel Out-of Round Diagnosis Based on Morphological Filtering and CEEMDAN-WVD

现有列车车轮失圆监测方法的准确性受车速及线路条件影响较大,为了更准确地监测车轮服役状态,文中提出基于形态学滤波和CEEMDAN-WVD的车轮失圆诊断方法:车辆轴箱垂向振动加速度经数学形态学滤波器滤波降噪后,运用完全噪声辅助聚合经验模态分解(CEEMDAN)将其分解为一系列的固有模态函数(IMF),然后选取能量熵增量相对较大的几阶IMF分量进行Wigner-Ville分布(WVD)计算,从而叠加得到轴箱振动加速度的多尺度时频图,最后根据多尺度时频图的分布特征来诊断车轮状态。通过仿真分析和工程实例研究结果表明,运用该方法可有效地识别复杂工况下的车轮服役状态。

The accuracy of the existing train wheel out-of-round monitoring methods is greatly affected by its speed and line conditions.In order to monitor the wheel service status more accurately,this paper proposes a wheel out-of-round diagnosis method based on morphological filtering and CEEMDAN-WVD:the vertical vibration acceleration of the axle-box of the vehicle is filtered by the mathematical morphology filter to reduce the noise,and then it is decomposed into a series of intrinsic modal functions(IMF)by the use of the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN).After decomposition into a series of intrinsic modal functions(IMF),the IMF components with relatively large energy entropy increments are selected for Wigner-ville distribution(WVD)calculation,so as to superimpose the multi-scale time-frequency diagrams of axle-box vibration acceleration.Finally,the distribution characteristics of the multi-scale time-frequency map are used to diagnose the wheel condition.The results of simulation analysis and engineering examples show that this method can be used to effectively identify the service state of wheels under complex working conditions.