考虑拓扑变化的电力系统暂态稳定评估  

Transient Stability Assessment of Power System Considering Change in Network Topology

作　　者：李欣[1] 刘晨凯 郭攀锋[1] 宁静 柳圣池 夏家辉 LI Xin;LIU Chenkai;GUO Panfeng;NING Jing;LIU Shengchi;XIA Jiahui(College of Electrical Engineering and New Energy,China Three Gorges University,Yichang 443002,China;State Grid Xiangyang Power Supply Company,Xiangyang 441000,China)

机构地区：[1]三峡大学电气与新能源学院,湖北宜昌443002 [2]国网湖北省电力有限公司襄阳供电公司,湖北襄阳441000

出　　处：《智慧电力》2021年第12期59-65,共7页Smart Power

基　　金：国家自然科学基金资助项目(52107107);电力系统智能运行与安全防御宜昌市重点实验室开放基金资助项目(2020DLXY04)。

摘　　要：针对电力系统拓扑频繁变化导致暂态稳定评估(TSA)困难的问题,提出一种新颖的基于深度学习的TSA方案。基于深度森林(DF)构建暂态稳定评估模型,创新性地提出考虑主动学习技术和分级策略的DF更新方案。该方案从样本生成和模型训练两方面最大限度节省更新时间,并在电力系统拓扑变化后快速更新DF模型,有较强实用性。最后,在新英格兰39节点系统中对所提TSA方案进行验证,结果表明该方案对拓扑频繁变化的电力系统有更强的适应性与鲁棒性。In view of the fact that it is difficult to conduct transient stability assessment(TSA)due to frequent change in the topology of power system,the paper proposes a novel TSA scheme based on deep learning.In addition,the paper establishes a transient stability assessment model based on deep forest(DF),and puts forward a DF updating scheme for the first time by considering active learning technology and grading strategy.The scheme saves the update time to an extreme according to sample generation and model training,updating the DF scheme quickly after the change in the topology.Finally,the TSA scheme is verified in New England 39-bus system,showing that the scheme has strong adaptability and robustness to the system with the frequent change in topology.

关 键 词：拓扑变化 暂态稳定评估 深度学习 深度森林 主动学习 

分 类 号：TM712[电气工程—电力系统及自动化]