基于改进小波系数奇异值分解和小波去噪的低频振荡时变-电网技术.pdf

第 36 卷 第 6 期 电 网 技 术 Vol. 36 No. 6 2012 年 6 月 Power System Technology Jun. 2012 文章编号：1000-3673 （2012 ）06-0141-07 中图分类号：TM 71 文献标志码：A 学科代码：470·40 基于改进小波系数奇异值分解和小波去噪的 低频振荡时变模式辨识 栾某德，刘涤尘，廖清芬，董超，欧阳利平 （武汉大学电气工程学院，湖北省 武汉市 430072 ） A Method to Identify Time-Varying Mode of Low Frequency Oscillation by Continuous Wavelet Transform Based on Raising Singular Value Decomposition of Wavelet Coefficient and Wavelet Denoising LUAN Moude, LIU Dichen, LIAO Qingfen, DONG Chao, OUYANG Liping (School of Electrical Engineering, Wuhan University, Wuhan 430072, Hubei Province, China) ABSTRACT ：Based on the combination of continuous wavelet 叉难以提取频率的情况，根据提升的小波系数奇异值分解频 transform (CWT) with singular value decomposition (SVD), a 率向量识别各阶振荡模式的频率。同时选用小波能量系数来 new algorithm to identify oscillation frequency of signal, 识别主导振荡模式，用小波软阈值去噪和 SVD 分解后矩阵 extract mode information and denoise signal by raising SVD of 重构来进行信号去噪。CWT 可以处理含时变振荡模式的低 wavelet coefficient is proposed. The condition that under high 频振荡信号，且对模式参数具有较高的辨识精度。仿真算例 noise level or closely spaced mode of noise, the wavelet ridges 验证了算法的有效性和适用性。 are unsharp and even the frequency is hard to extract due to the 关键词：连续小波变换(CWT) ；奇异值分解(SVD) ；时变振 aliasing and intersection of wavelet ridges can be overcome by 荡；小波能量系数；主导模式；小波软阈值去噪 the proposed method, and the frequencies of oscillation modes in different orders can be identified according to frequency 0 引言 vectors of the raised SVD of wavelet coefficients. Meanwhile 随着电网互联，负荷加重、快速励磁系统的使 the wavelet energy coefficient is chosen to identify the 用，电力系统的结构和