基于短时傅立叶变换的电能质量扰动识别与采用奇异值分解的扰动时间定位.pdfVIP

第 35 卷 第 8 期 电 网 技 术 Vol. 35 No. 8 2011 年 8 月 Power System Technology Aug. 2011 文章编号：1000-3673 （2011 ）08-0174-07 中图分类号：TM 711 文献标志码：A 学科代码：470·4054 基于短时傅里叶变换的电能质量扰动识别 与采用奇异值分解的扰动时间定位 徐永海，赵燕 （华北电力大学电气与电子工程学院，北京市 昌平区 102206 ） Identification of Power Quality Disturbance Based on Short-Term Fourier Transform and Disturbance Time Orientation by Singular Value Decomposition XU Yonghai, ZHAO Yan (Department of Electrical and Electronic Engineering, North China Electric Power University, Changping District, Beijing 102206, China) ABSTRACT: It is proposed to identify power quality 摘要：提出了用短时傅里叶变换作为时频信号分析工具研究 disturbances by using short-term Fourier transform as the 电能质量扰动识别问题，同时提出了用奇异值分解技术研究 analysis tool of time-domain signals, meanwhile the problem to 扰动时间定位问题。从扰动电压信号短时傅里叶变换后得到 research the time orientation of disturbance by singular value 的二维频谱幅值矩阵中提取出4 个特征序列，生成频谱峰值 decomposition is put forward. Four feature sequences are 曲线、基频幅值曲线、高频幅值曲线和幅值标准差曲线， extracted from the two-dimensional frequency spectrum 这些曲线用 6 个特征量来表征。当 6 个特征量中的某几个 amplitude matrix obtained from the short-term Fourier 满足一定的取值组合时能够唯一确定一种扰动，文中通过 transform of disturbance voltage signals to create four curves, 建立决策树，实现多种单一与复合扰动的识别。利用采样 i.e., the curve of spectrum peak values, the curve of 信号构造 Hankel 矩阵，对此矩阵进行奇异值分解，通过分 fundamental frequency amplitudes, the curve of high-frequency 量信号的构造并从中提取模极大值点，进行扰动时间定 amplitudes and the standard deviation curve of amplitudes, and 位。仿真结果表明，本文提出的方法能够实现 8 种单一与 these curves are characterized by six