基于经验模式分解(EMD)的齿轮箱齿轮故障诊断技术研究

简述了齿轮箱传统信号分析技术与经验模式分解(EMD)技术的异同，并详细论述了EMD的分解原理和富立叶变换的关系。针对齿轮箱振动加速度数据，运用EMD分解技术，得到IMF(Intrinsic Mode Function)模式分量，提出了基于IMF及功率谱(PSD)的IMFPSD216指标和基于IMF及滤波统计的IMFFLT指标，并验证了它们的有效性。运用这些指标正确明显地分辨出齿轮箱齿轮失效。

Study on Fault Diagnosis of Gears Using Empirical Mode Decomposition

Vibration of a gearbox is cumulated nonlinearly by a number of vibration sources, which transmit in the form of dispersion and reverberation. In order to recognize the generic vibration mode of the component, this paper discusses differences between traditional signal processing technique and empirical mode decomposition (EMD) for a gearbox, and details in the theory of EMD and its mathematical and physical significance. Using vibration acceleration data sets of the gearbox, intrinsic mode functions (IMFs) and their power spectrum density(PSD) extracted with EMD and Fourier analystical techniques. Furthermore, the effective gear fault indexes of the IMF and PSD on 216 frequency band (IMFPSD216) based on the 2nd IMF components(IMF2) and its PSD, and the IMF using filter (IMFFLT) based on IMF2 and its filtered statistical signal, are given. A successful diagnosis of the gear fault is obtained using these indexes.