基于频域分解法和广域稳态量测的低频振荡模式辨识  

Identification of low-frequency oscillation modes based on frequency domain decomposition method and wide area ambient measurements

作　　者：陈刚 吴小辰[3] 柳勇军[1] 李鹏 廖瑞金[2] 王予疆[2] 何潜[4] 

机构地区：[1]南方电网科学研究院,广东广州510080 [2]输配电装备及系统安全与新技术国家重点实验室(重庆大学),重庆400044 [3]南方电网电力调度通信中心,广东广州510623 [4]重庆电力调度控制中心(重庆电力公司),重庆400014

出　　处：《电力系统保护与控制》2011年第12期1-6,21,共7页Power System Protection and Control

基　　金：国家重点基础研究发展计划项目(973计划项目)(2009CB724505-1);中国南方电网有限责任公司科技专项(K0911)

摘　　要：对广域稳态量测信号将频域分解方法引入电网低频振荡模式辨识并进行了改进完善。推导了系统的频响函数、输出功率谱密度矩阵、特征量之间的关系,给出了弱阻尼模式下功率谱密度矩阵的最大奇异值与特征量之间的关系表达式;采用模态幅值相干系数以确定奇异值曲线谱峰附近单模态区域大小,最后采用最小二乘方法求取准确的模式频率和阻尼系数。线性时不变系统时域仿真、WECC9节点系统时域仿真以及实测PMU数据分析表明,所提方法适用于弱阻尼振荡模式辨识,对噪声具有较强鲁棒性,是一种极具实用价值的在线应用方法。Frequency Domain Decomposition（FDD） is introduced to identify low-frequency oscillation modes by analysis of steady-state ambient measurements in power system and some modifications to FDD are made.The relation expression between maximum singular value of PSD matrix and system eigenvalue under poorly damped mode is presented by detailed derivation of the relations among frequency response function（FRF）,output power spectral density（PSD）,and system eigenvalues.Modal Amplitude Coherence（MAC） is used to determine mode area size in the neighborhood of peak point on maximum SVD curve.Finally,the exact modal frequency and damping ratio can be estimated using least square method.The technique is illustrated on linear time-invariant system,WECC 9-bus system and ambient PMU measurements,to show its effectiveness.The suggested method is suitable for identification of poorly damped oscillatory modes,and is robust to noise,and has important value for real-time application.

关 键 词：广域稳态量测 低频振荡 模式辨识 频域分解方法 奇异值分解 

分 类 号：TM712[电气工程—电力系统及自动化]