极高光伏渗透率下基于潮流雅可比矩阵和卷积神经网络的静态电压稳定在线预测  

Power Flow Jacobian Matrix Based Static Voltage Stability Forecast by CNN Considering Extremely High PV Penetration

作　　者：吴倩红[1] 韩蓓[1] 李国杰[1] 汪可友[1] WU Qianhong;HAN Bei;LI Guojie;WANG Keyou(MOE Key Lab of Power Transmission and Power Conversion and Control(Shanghai Jiao Tong University),Minhang District,Shanghai 200240,China)

机构地区：[1]电力传输与功率变换控制教育部重点实验室(上海交通大学),上海市闵行区200240

出　　处：《中国电机工程学报》2021年第12期4058-4067,共10页Proceedings of the CSEE

基　　金：国家自然科学基金项目(51877133,51477098)。

摘　　要：大量光伏接入电力系统,给系统的电压稳定带来了挑战。该文针对极高光伏渗透率,即瞬时光伏渗透率可能大于100%的情况,分析电压会产生崩溃的现象,提出双向静态电压稳定裕度(voltage stability margin,VSM)的概念。由于潮流雅可比矩阵具有天然的网格结构性、拓扑变化性与电压相关性,而卷积神经网络具有强的学习能力、泛化能力,且可以专门处理网格结构数据,因此,潮流雅可比矩阵作为输入的卷积神经网络模型能够预测双向VSM。通过IEEE-33节点系统与IEEE-123节点系统对所提方法进行验证,结果显示,所提方法能够在系统网络拓扑结构发生变化时保持较高的预测精度,且具有较强的泛化能力。The expansion of photovoltaic(PV) generations in power system brings challenges to system voltage stability. Regarding the extremely high PV penetration, this paper analyzed the voltage collapse and proposed bidirectional voltage stability margin(VSM). Power flow Jacobian matrix is naturally grid-like, topology-involved and has voltage-relevant data, whereas convolutional neural network(CNN) provides deep model architecture with strong capability of learning, generalization and processing grid-like topology data. This paper utilized power flow Jacobian matrix as input and forecasted bidirectional VSM by CNN. The validity of the proposed method was verified through the IEEE 33-bus system and IEEE 123-node test feeder. The results show that the proposed method maintains high accuracy when system topology changes, and has stronger generalization ability.

关 键 词：极高光伏渗透率 电压稳定裕度 潮流雅可比矩阵 卷积神经网络 拓扑结构 

分 类 号：TM734[电气工程—电力系统及自动化]