中文摘要：

提出了基于局部均值分解（LMD）的同步电机参数辨识方法。采用LMD从短路电流中提取直流电流和基波电流,然后分别采用稳健回归最小二乘和Prony算法对直流电流和基波电流进行辨识,进而获得同步电机参数。以理想突然短路电流信号为例,通过仿真分析了滑动平均跨度与LMD循环次数和电流相对均方误差的关系,确定了滑动平均跨度。高信噪比（30 d B）时,由于LMD具有平滑滤波功能而无模态混叠现象发生。低信噪比（15 d B）时,提出了基于前置滑动平均LMD的短路电流分离方法,可有效获取直流电流和基波电流分量。较之经验模态分解（EMD）,基于LMD的理想突然短路电流分解效果更好。仿真结果表明,与EMD方法相比,所提方法受噪声影响较小,参数辨识精度更高。

英文摘要：

A method of synchronous motor parameter identification based on the LMD（Local Mean Decomposition） is proposed,which extracts the DC current and fundamental current from the short circuit current by LMD and identifies them respectively by the least squares regression and Prony algorithm to obtain the synchronous motor parameters. With the ideal sudden short circuit current signal as an example,the relationship between the LMD moving average span and the LMD cycles and that between LMD moving average span and the relative mean square error of short circuit current are analyzed to determine the moving average span. When SNR（Signal-to-Noise Ratio） is high（30 d B）,there is no modal aliasing due to the smooth filtering function of LMD. When SNR is low（15 d B）,the short circuit current separation method based on the pre-moving average LMD is proposed to effectively extract DC current and fundamental current.Compared with the EMD（Empirical Mode Decomposition）,the ideal sudden short circuit current decomposition effect of LMD is better,and the simulative results also show that,the proposed method is less influenced by noise and has higher parameter identification precision.