基于动态模式分解的电力系统主导振荡模式及参数识别方法研究  

Extracting dominant electromechanical oscillation modes using dynamic mode decomposition

作　　者：梁钰 刘红岩 毛李帆 万信书 林明健 LIANG Yu;LIU Hong⁃yan;MAO Li⁃fan;WAN Xin⁃shu;LIN Ming⁃jian(Electric Power Research Institute of Hainan Power Grid Co.,Ltd.,Haikou 570311,China;Hainan Key Laboratory of Physical and Chemical Analysis,Haikou 570311,China;Hainan Electric Power Dispatching Centre,Haikou 570203,China)

机构地区：[1]海南电网有限责任公司电力科学研究院,海南海口570311 [2]海南省理化分析重点实验室,海南海口570311 [3]海南电力调度中心,海南海口570203

出　　处：《电子设计工程》2020年第2期139-143,共5页Electronic Design Engineering

基　　金：中国南方电网有限责任公司科技项目（073000KK52180003)

摘　　要：本文以广域同步量测数据为基础,提出了一种基于动态模式分解的电力系统机电振荡模态参数提取与分析方法。动态模式分解算法是一种数据驱动的模态分析手段,可以准确捕捉电力系统机电振荡模态信息,并建立相应的动力学降阶模型,能够实现数据驱动下的机电振荡模态、振荡频率及阻尼比的识别。同时,本方法能够准确获取各机电振荡模式的能量幅值,并以模式能量为依据提取主导振荡模式及相应的动态趋势(频率和阻尼比)和空间关系(振型)信息。文中利用IEEE 16机68节点系统进行数值仿真计算与分析,结果证明了所提出方法的有效性。This paper proposes an improved dynamic mode decomposition(DMD)algorithm to extract electromechanical oscillation modes from the noise-containing data of wide-area measurements.This algorithm effectively improves the ability of the dynamic mode decomposition(DMD)technique to resist noise in the observed data by using variable projection and a finite-difference style approximation method.The proposed method provides a reliable extraction of the spatial relationships(mode shape),dynamic trends(frequency and damping ratio)and dominant modes using the derived energy relationships in the measured data.The performance of the proposed method has been investigated using simulated data from an IEEE 16-generator system.

关 键 词：机电振荡 主导模式 模态参数 动态模式分解 

分 类 号：TM721[电气工程—电力系统及自动化]