基于UBSS算法的电力系统低频振荡辨识方法  

Identification of Low-frequency Oscillation in Power Systems Based on Underdetermined Blind Source Separation Algorithm

作　　者：夏远洋 李啸骢 徐俊华[1] 刘治理 刘源 XIA Yuanyang;LI Xiaocong;XU Junhua;LIU Zhii;LIU Yuan(Guangxi Key Laboratory of Power System Optimization and Energy Technology(Guangxi University),Nanning 530004,Guangxi Zhuang Autonomous Region,China;Intelligent Manufacturing College,Nanning University,Nanning 530200,Guangxi Zhuang Autonomous Region,China;Yalong River Hydropower Development Co.,Ltd.,Chengdu 610051,Sichuan Province,China)

机构地区：[1]广西电力系统最优化与节能技术重点试验室(广西大学),广西壮族自治区南宁市530004 [2]南宁学院智能制造学院,广西壮族自治区南宁市530200 [3]雅砻江流域水电开发有限公司,四川省成都市610051

出　　处：《中国电机工程学报》2024年第13期5073-5083,I0005,共12页Proceedings of the CSEE

基　　金：国家自然科学基金项目(51267001,U1965202);广西自然科学基金项目(2014GXNSFAA118338);广西科学研究与技术开发计划项目(14122006-29)。

摘　　要：低频振荡监测和分析对电力系统故障诊断和电网恢复至关重要。该文提出一种基于欠定盲源分离原理的低频振荡模式辨识方法,包括欠定盲源分离(underdetermined blind source separation,UBSS)和希尔伯特变换(Hilbert transform,HT)。首次系统地提出并论证含欠定盲源分离、模式定阶和振荡参数的辨识方法。提出的UBSS-HT方法利用能量比函数确定故障时刻,利用贝叶斯信息准则(Bayesian information criterion,BIC)实现模式定阶,阐述维度空间理论,论证构建虚拟多通道的可行性,通过盲源分离来实现源信号分离,最后通过HT在希尔伯特空间来辨识振荡参数。通过大量的系统建模仿真和现场录波数据试验评估所提方法的性能,验证该方法的有效性、准确性和抗干扰能力。Low frequency oscillation(LFO)monitoring and analysis are crucial for power system fault diagnosis and grid recovery.In this paper,LFO mode identification method based on underdetermined blind source separation principle is proposed,including underdetermined blind source separation(UBSS)and Hilbert transform(HT).A solution that includes underdetermined blind source separation,modal order determination,and oscillation parameter identification has been proposed and demonstrated for the first time in a systematic manner.The proposed UBSS-HT method uses the energy ratio function to determine the fault time,and uses Bayesian information criteria(BIC)to achieve modal order determination.It expounds the theory of dimension space,demonstrates the feasibility of building virtual multi-channel,realizes source signal separation through blind source separation,and finally identifies oscillation parameters through HT in Hilbert space.The performance of the proposed method is evaluated through a large number of system modeling simulations and on-site recording data experiments,verifying its effectiveness,accuracy,and anti-interference ability.

关 键 词：欠定盲源分离 低频振荡 能量比函数 维度变换 源数估计 

分 类 号：TM743[电气工程—电力系统及自动化]