基于CMF-EEMD的风电齿轮箱多故障特征提取

摘要

针对EMD( empirical mode decomposition)模态混叠现象和由于所添加白噪声幅值单一而影响EEMD( ensemble empirical mode decomposition)分解精度等问题,提出了一种新的信号处理方法CMF-EEMD。 CMF( combined mode function)将EMD分解得到敏感的IMFs按高低频进行组合,形成两个包含高低频的本征模态函数Ch 和CL,然后通过添加不同的白噪声幅值对Ch 和CL 分别进行EEMD分解,最后对敏感的IMFs进行循环自相关函数解调分析。将提出方法应用于仿真信号和风力齿轮箱试验台的振动信号,成功提取了多故障特征频率,验证了此方法的有效性。并通过与添加单一白噪声幅值进行对比分析,凸显此方法具有更高的分解精度。%In view of the problems such as empirical mode decomposition( EMD) modal aliasing phenome-non and ensemble empirical mode decomposition(EEMD)precision which affected by the singularity of amplitude of the added white noise, an improved EEMD with combined mode function( CMF) was pro-posed. Combined mode function( CMF) was used as the pre-filter to improve EEMD decomposition re-sults. CMF is combining the neighboring intrinsic mode functions ( IMFs) which are obtained by EMD to get two new IMFs Ch and CL. Ch contains high frequency components and CL contains low frequency com-ponents. The proper added noise amplitude was determined according to the vibration characteristics to decompose Ch and CL with EEMD, and the purpose is that EEMD is further improved to increase the ac-curacy and effectiveness of its decomposition results. Finally, what extracts weak fault frequency more ef-fectively is cyclic autocorrelation function analysis for every characteristic IMF. The proposed method is applied to analyze the multi-fault of a wind power growth gearbox setup, and the results confirm the ad-vantage of the proposed method over EEMD with cyclic autocorrelation function.