基于经验模态分解谱峭度重构峰值定位的电机噪声溯源方法

Motor Noise Tracking Method Based on EMD-KR-FP

针对无刷直流电机噪声溯源问题,提出了一种基于经验模态分解谱峭度重构峰值定位(EMD-KR-FP)的电机噪声溯源方法。首先经理论计算得到径向电磁力、转矩脉动和共振引起的电磁噪声的特征频率,定义特征频率集,并采用经验模态分解(EMD)方法对电机噪声时域信号进行分解;然后根据谱峭度理论筛选本征模态函数(IMF)分量进行信号重构,对重构信号进行傅里叶变换,使用峰值定位算法对频谱上贡献最大的几个峰值进行峰值频率提取并与特征频率集的频率进行对照,确定引起电机电磁噪声的原因。试验结果表明,该方法有效,与传统阶次分析法相比,工作量显著降低。

A motor noise tracking method based on Empirical Mode Decomposition spectral Kurtosis Reconstruction Find Peaks(EMD-KR-FP)was proposed to track the noise of brushless DC motor.Firstly,the characteristic frequencies of radial electromagnetic force,torque ripple and resonance induced electromagnetic noise were obtained by theoretical calculation.The characteristic frequency set was defined,the time-domain signal of motor noise was decomposed by Empirical Mode Decomposition(EMD)method.Then,according to the spectral Kurtosis theory,the Intrinsic Mode Function(IMF)component was screened for signal reconstruction,the reconstructed signal was processed by Fourier transform.Peak location algorithm was used to extract the peak frequency of several peaks with the largest contribution on the spectrum and compare with the frequency of characteristic frequency set,so as to determine the cause of electromagnetic noise of the motor.Test results show that this method is effective and reduces the workload of traditional order analysis method.