摘要
当大量逆变器接入电网时,会发生谐波不稳定问题。并网逆变器dq阻抗的准确获取是分析该问题的关键前提之一。仅基于知识驱动的白箱建模方法难以适应工程现场信息保密的实际情况,黑箱或灰箱测量方法会向实际系统持续注入谐波扰动;仅基于数据驱动的方法难以合理确定输入变量。针对这些缺陷,提出了一种基于知识和数据联合驱动的并网逆变器多工况dq阻抗获取方法。首先,基于知识驱动建立小信号模型以选取输入变量;然后,结合数据驱动具有较好拟合输入输出变量间非线性关系的优点,采用基于遗传算法的核极限学习机获取多工况dq阻抗;最后,借助RT-LAB搭建半实物实时仿真平台,验证所提方法的有效性及优越性。
When a large number of inverters are connected to power grids,the problem of harmonic instability will occur. Accurate dq impedance acquisition of grid-connected inverters is one of the key preconditions to analyze the problem. The knowledge-driven white-box modeling methods are difficult to adapt to the actual situation that the information in the engineering field is confidential,but also the black box or grey box measurement methods will continually inject the harmonic disturbances into the actual power system, while the data-driven based methods are hard to determine the input variables reasonably. Aiming at these deficiencies, a combined data and knowledge-driven impedance acquisition method for grid-connected inverter under multiple operating conditions is proposed. Firstly, the knowledge-driven small-signal models are established to pick the input variables. Secondly, combined with the advantages that data-driven can better fit the nonlinear relationship between input and output variables, the kernel extreme learning machine is optimized by the genetic algorithm to obtain the dq impedance under multiple operating conditions. Finally,with the help of RT-LAB, a semi-physical real-time simulation platform is built to verify the effectiveness and superiority of the proposed method.
作者
吴滨源
李建文
李永刚
王月
刘淇玉
杨夷南
WU Binyuan;LI Jianwen;LI Yonggang;WANG Yue;LIU Qiyu;YANG Yinan(State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources(North China Electric Power University),Baoding 071003,China;Feidong Power Supply Company of State Grid Anhui Electric Power Supply Co.,Ltd.,Hefei 231600,China)
出处
《电力系统自动化》
EI
CSCD
北大核心
2022年第6期83-91,共9页
Automation of Electric Power Systems
基金
国家电网公司科技项目(SGJBTS00FZJS2100341)。
关键词
数据驱动
知识驱动
并网逆变器
阻抗
遗传算法
核极限学习机
谐波稳定性
data-driven
knowledge-driven
grid-connected inverter
impedance
genetic algorithm
kernel extreme learning machine
harmonic stability
