基于局部均值分解和归一化最小均方的宽频振荡检测方法  

Broadband Oscillation Signal Detection Method Based on Local Mean Decomposition andNormalized Least Mean Square

作　　者：吴为 曾德辉 聂欣昊 卢家俊 李超杰 WU Wei;ZENG Dehui;NIE Xinhao;LU Jiajun;LI Chaojie(Electric Power Research Institute,China Southern Power Grid,Guangzhou 510080,China;Guangzhou Jiayuan Electric Power Technology Co.,Ltd,Guangzhou 510610,China)

机构地区：[1]南方电网科学研究院,广州510080 [2]广州嘉缘电力科技有限公司,广州510610

出　　处：《电力系统及其自动化学报》2024年第3期48-58,共11页Proceedings of the CSU-EPSA

基　　金：广东省基础与应用基础研究基金资助项目(2021A1515012602)。

摘　　要：为精确采集和分析电力系统宽频振荡信号,提出基于改进局部均值分解与归一化最小均方算法相结合的宽频振荡检测新方法。首先利用归一化最小均方算法对宽频振荡信号进行降噪处理,进而采用改进局部均值分解算法提取该降噪信号的乘积函数;然后对上述乘积函数分量进行希尔伯特变换,求解信号瞬时频率的高频突变点,实现对振荡起止时刻的准确定位。仿真实验表明,本文所提方法能准确求解宽频振荡信号,且在强噪声下仍具有很高的精度。To accurately collect and analyze the broadband oscillation signal of power system,a novel method for broadband oscillation signal detection is put forward,which is based on the combination of the improved local mean decomposition(ILMD)and normalized least mean square(NLMS)algorithms.First,the NLMS algorithm is used to reduce the noise in the broadband oscillation signal.Second,the ILMD algorithm is used to extract the product function(PF)of the noise-reduced signal.Third,Hilbert Transform is performed on the PF component to solve the high-frequency mutation of the signal’s instantaneous frequency,so as to achieve an accurate localization of the starting and stopping moments of the oscillation.Simulation results show that the proposed method can accurately solve the broadband oscillation signal,and it still has a high accuracy under strong noise.

关 键 词：强噪声干扰 归一化最小均方 局部均值分解 宽频振荡 希尔伯特变换 

分 类 号：TM712[电气工程—电力系统及自动化]