| 引用本文: | 肖先勇,刘子菁,马晓阳,等.基于矫正窗函数IFFT和最小二乘法的宽频振荡检测方法[J].电力系统保护与控制,2025,53(24):65-77.[点击复制] |
| XIAO Xianyong,LIU Zijing,MA Xiaoyang,et al.Wideband oscillation detection method based on window-adjusted IFFT and least squares algorithm[J].Power System Protection and Control,2025,53(24):65-77[点击复制] |
|
| 摘要: |
| 准确估计电力系统中的宽频振荡参数能够为振荡抑制提供技术支撑。然而，由于宽频振荡与谐波、间谐波的频率范围相近，在将现有的次/超同步振荡估计扩展到宽频振荡估计时，除了需要克服来自基波的频谱泄漏，还要考虑宽频振荡的耦合特性和阻尼特性。对此，提出了一种基于矫正窗函数的插值快速傅里叶变换(interpolated fast Fourier transform, IFFT)和最小二乘(least squares, LS)法的宽频振荡检测方法。首先，将窗函数经旋转因子矫正，并通过相邻谱线矢量相消方式，插值获取频率值。然后，根据振荡特性，通过峰值提取函数，求解成分数目并识别振荡成分。最后，通过泰勒级数逼近振荡并求解伪逆矩阵，得到LS法修正后的结果。仿真结果表明，所提出的方法能够克服多成分干扰，实现4 Hz至2.4 kHz频率范围内随机宽频振荡参数的精确估计。 |
| 关键词: 宽频振荡 插值快速傅里叶变换 信号检测 最小二乘 阻尼因子 |
| DOI：10.19783/j.cnki.pspc.250102 |
| 投稿时间：2025-01-24修订日期：2025-06-13 |
| 基金项目:国家电网有限公司总部管理科技项目资助(5100- 202312421A-3-2-ZN) |
|
| Wideband oscillation detection method based on window-adjusted IFFT and least squares algorithm |
| XIAO Xianyong1,LIU Zijing1,MA Xiaoyang1,YUAN Zehui1,E Zhijun2,LI Shupeng2,LIU Tao2 |
| (1. College of Electrical Engineering, Sichuan University, Chengdu 610065, China;
2. State Grid Tianjin Electric Power Company, Tianjin 300010, China) |
| Abstract: |
| Accurate estimation of wideband oscillation parameters in power systems can provide essential technical support for oscillation suppression. However, because wideband oscillations share similar frequency ranges with harmonics and interharmonics, extending existing subsynchronous/supersynchronous oscillation estimation to wideband oscillation estimation requires not only overcoming spectrum leakage from the fundamental frequency, but also addressing the coupling and damping characteristics of wideband oscillations. To this end, a window-adjusted interpolated fast Fourier transform (IFFT) and least squares (LS) method are proposed for wideband oscillation estimation. First, the window function is adjusted using a rotation factor, and adjacent spectral-line vector cancellation is applied to obtain interpolated frequency values. Then, based on oscillation characteristics, a peak-extraction function is used to determine the number of components and identify oscillatory components. Finally, LS corrected results are obtained by Taylor series approximation and pseudoinverse matrix solving. Simulation results show that the proposed method can effectively overcome multi-component interference and achieve accurate estimation of random wideband oscillation parameters in the 4 Hz to 2.4 kHz frequency range. |
| Key words: wideband oscillation interpolated fast Fourier transform signal measurement least squares damping factor |
