发电技术 ›› 2022, Vol. 43 ›› Issue (1): 19-31.DOI: 10.12096/j.2096-4528.pgt.21083

• 智能电网 • 上一篇    下一篇

基于类噪声数据的电力系统低频振荡模态参数辨识

闫红艳1,2, Kwon Hwang Jin2, 高艳丰1   

  1. 1.河北工程大学水利水电学院,河北省 邯郸市 056038
    2.韩国又石大学能源工程系,韩国 镇川 365-803
  • 收稿日期:2021-06-18 出版日期:2022-02-28 发布日期:2022-03-18
  • 作者简介:闫红艳(1980),女,博士研究生,讲师,研究方向为电力系统稳定性分析与控制,yhyan118@126.com
    Hwang Jin Kwon(1961),男,教授,博士生导师,研究方向为控制与仪表,电力系统稳定性和智能电网;
    高艳丰(1979),男,副教授,硕士生导师,研究方向为电气设备状态监测与故障诊断,电力系统保护与控制,本文通信作者, gaoyanfeng01@126.com
  • 基金资助:
    河北省社会科学发展研究课题(20210201316);河北省高等学校科学技术研究项目(ZD2021021)

Modal Parameter Identification of Low Frequency Oscillation in Power System Based on Ambient Data

Hongyan YAN1,2, Jin Kwon HWANG2, Yanfeng GAO1   

  1. 1.School of Water Conservancy and Hydroelectric Power, Hebei University of Engineering, Handan 056038, Hebei Province, China
    2.Department of Energy Engineering, Woosuk University, Jinchon-gun 365-803, Republic of Korea
  • Received:2021-06-18 Published:2022-02-28 Online:2022-03-18
  • Supported by:
    Social Science Development Research Project of Hebei Province(20210201316);Science and Technology Research Project of Higher Education of Hebei Province(ZD2021021)

摘要:

低频振荡是影响互联电力系统安全稳定运行的关键问题之一,提出采用变分模态分解(variational mode decomposition,VMD)提取类噪声数据的低频振荡信号,基于离散傅里叶变换(discrete Fourier transform,DFT)曲线拟合的电力系统低频振荡模态辨识方法。首先,采用VMD分解滤除类噪声数据信号中的直流分量,提取出低频振荡信号,利用模态相关系数确定VMD分解个数,提高了信号分解的时效性;其次,建立类噪声数据自回归滑动平均(auto regressive moving average,ARMA)数学模型,模拟产生数据信号,利用低频振荡信号自相关函数的DFT曲线拟合估计拉普拉斯变换系数,提取机电振荡特征参数;最后,采用模拟数据和某实测相量测量单元数据验证了该方法的可行性和有效性。试验表明,采用VMD算法和基于DFT的曲线拟合法提取低频振荡特征参数,有效提高了机电小干扰稳定评估的实时性。

关键词: 低频振荡, 类噪声数据, 自回归滑动平均(ARMA)模型, 变分模态分解(VMD), 离散傅里叶变换(DFT)曲线拟合

Abstract:

Low frequency oscillation is one of the key problem affecting the safe and stable operation of interconnected power system, variational modal decomposition (VMD) was used to extract low-frequency oscillation signals from ambient data, and a method of low-frequency oscillation modal identification for power systems discrete Fourier transform (DFT)-based curve fitting was proposed in this paper. Firstly, the DC component of ambient data signals was filtered by VMD decomposition to extract low-frequency oscillation signals. The number of VMD decomposition was determined by modal correlation coefficient, which improved the timeliness of signal decomposition. Secondly, the auto regressive moving average (ARMA) model of ambient data was established to simulate the generation of data signals. The DFT curve fitting of low-frequency oscillation signal autocorrelation function was used to estimate the Laplace transform coefficient and extract characteristic parameters of electromechanical oscillation. Finally, Simulation data and some measured phasor measurement unit (PMU) data are used to verify the feasibility and effectiveness of the method. The experiment shows that the sampling VMD algorithm and the curve fitting method based on DFT can extract the characteristic parameters of low-frequency oscillation, which effectively improves the real-time performance of electromechanical small interference stability.

Key words: low frequency oscillation, ambient data, auto regressive moving average (ARMA) model, variational modal decomposition (VMD), discrete Fourier transform (DFT) curve-fitting

中图分类号: