摘要
为防止因输电线路故障导致事故影响的范围进一步扩大,提出一种基于改进的长短期记忆网络GRU对电力系统中输电线路的故障在线识别的方法。采用固定时间窗口移位标记的方法对故障起始时刻附近数据编码,搭建故障起始时刻判别逻辑;再根据返回的起始时刻对故障清除前的节点运行数据映射故障位置及类型。最后在PSCAD/EMTDC中以新英格兰10机39节点为例利用python API接口实现联合仿真。结果表明该方法能够快速准确地识别故障,模型准确率在98.91%以上,可为后续动作和故障分析提供依据。
In order to prevent further expansion of impact scope of the accident resulted from transmission line failure,an improved long-short-term-memory network GRU for on-line fault identification of transmission lines in power system is proposed.The method of fixed time window shift marking is used to encode the data near the fault start time,and the fault start time discrimination logic is built;then according to the returned start time,the node operation data before fault clearing is mapped to the fault location and type.Finally,taking New England 10 machine 39 bus in PSCAD/EMTDC as an example,the co-simulation is realized by using Python API interface.The results show that the method can quickly and accurately identify the fault,and the accuracy of the model is more than 98.91%,which can provide the basis for the follow-up action and fault analysis.
作者
周红
王海云
ZHOU Hong;WANG Haiyun(Engineering Research Center of Ministry of Education for Renewable Energy Generation and Grid Connection Technology,Xinjiang University,Urumqi 830047,China)
出处
《智慧电力》
北大核心
2021年第8期55-62,103,共9页
Smart Power
基金
国家自然科学基金资助项目(52067020)
新疆维吾尔自治区重点研发任务(2020B02001)。
