针对局部均值分解（Local Mean Decomposition，LMD）算法应用于电能质量扰动检测时存在“端点效应”与滑动平均收敛速度慢，严重影响测量精度的问题，提出一种改进局部均值分解方法（Modified LMD，MLMD）。通过分段三次Hermite插值取代滑动平均法，有效改善LMD收敛慢、受平滑长度影响的弊端。为避免延拓长度不够而导致的“延拓失败”情形，在镜像延拓法的基础上结合“奇延拓”方法提出改进镜像延拓法。针对“直接法”求频率存在“毛刺现象”的弊端，文中改用希尔伯特变换（Hilbert Transform，HT）求取瞬时频率。最后，将MLMD分别应用于单一扰动信号与复合谐波信号的检测，相较传统的经验模态分解方法（Empirical Mode Decomposition，EMD），MLMD方法可有效抑制“端点效应”，同时能更准确的定位扰动信号的起止时刻，并且对高次谐波信号有更好的提取能力。
英文摘要:
Local mean decomposition (LMD) algorithm has the problems of “endpoint effect” and slow convergence speed of moving average (MA) algorithm which would affect detection accuracy seriously when used to detect power quality disturbance. To solve this problem, this paper presents a modified local mean decomposition (MLMD). Piecewise cubic Hermite interpolation is used to replace the MA method, which effectively improves the disadvantages of slow LMD convergence and the effect of smoothing length. In order to avoid "failure of extension" caused by insufficient extending length, combined with "odd extending" method, an improved mirror extending method is proposed based on the traditional mirror extending method. To solve the problem of the "burr phenomenon" in the "direct method", this paper adopts the Hilbert transform (HT) to obtain the instantaneous frequency. Finally, MLMD is applied to the detection of single disturbance signal and composite harmonic signal respectively. Compared with the traditional empirical mode decomposition method (EMD), the MLMD method can effectively suppress the "endpoint effect" and locate the starting and ending moments of disturbance signals more accurately, which has a better ability to extract high-order harmonic signals.
