改进的经验模式分解在机械故障诊断中的应用

针对经验模式分解(empirical mode decomposition,简称EMD)在工程应用中存在的端点效应和模式混叠问题,提出了一种改进的EMD方法。首先,利用遗传支持向量回归对短信号进行延拓;然后,采用改进的包络拟合方法并结合总体经验模式分解(ensemble empirical mode decomposition,简称EEMD)处理信号,数值仿真结果验证了该方法能够有效抑制端点效应和模式混叠;最后,利用该方法并结合包络解调对滚动轴承内圈故障信号进行实验与分析。结果表明,与EMD相比,该方法可以更有效地提取故障特征,满足机械设备故障诊断工程实际需求。

Applications of Improved Empirical Mode Decomposition in Machinery Fault Diagnosis

Aiming at the end effect and mode mixing of empirical mode decomposition (EMD) in practical applications, an improved EMD method consisting of the following techniques is proposedafter analyzing several solutions. First, short signals are extended with genetic support vector regression, then an ensemble EMD combined with an alternative envelope for the sifting procedure is employed to process the extended signals. The simulation results of two nonlinear and one simulated fault signals, as well as comparisons with other EMD methods, verify the capability of the proposed method to alleviatethe end effect and mode mixing. Applying the proposed method and envelope analysis to fault diagnosis of the ball bearing with an inner raceway defect, the results demonstrate the superiority of the method in extracting fault characteristic information for engineering applications.