基于深度学习的输电线路广域后备保护方法  

Wide-area back-up protection method for electric transmission line based on deep learning

作　　者：陈大华 王诚 符方友 CHEN Da-hua;WANG Cheng;FU Fang-you(Hainan Power Grid Co.,Ltd.,Power Dispatching Control Center,Haikou 570100,China;Hainan Power Grid Co.,Ltd.,Information and Communication Branch,Haikou 570100,China)

机构地区：[1]海南电网有限责任公司电力调度控制中心,海口570100 [2]海南电网有限责任公司信息通信分公司,海口570100

出　　处：《信息技术》2021年第10期65-69,共5页Information Technology

基　　金：海南电网科技项目(GZKJXM20170867)。

摘　　要：系统在突加负载或功率波动等压力条件下,距离继电器容易误判系统发生故障而误动。文中提出了一种基于深度学习的广域后备保护方法,通过建立状态评估和故障识别两种深度神经网络来进行保护决策。以相量测量单元(PMU)数据为输入,训练深度神经网络以进行决策逻辑的开发。系统在正常情况下,距离继电器会按常规工作,但在压力条件下,需进行故障识别,识别为故障情况距离继电器才动作。在9节点系统上对所提出的方法进行了测试,并与传统方案进行了比较。测试结果表明,该方法可显著减少距离继电器的误动,具有更高的保护效率。When the system is under pressure conditions such as sudden load or power fluctuations,distance relay is easy to misjudge the system failure,thus output maloperation.This paper proposes a wide-area back-up protection method based on deep learning,which makes protection decisions by establishing two deep neural networks for state assessment and fault identification.Taking phasor measurement unit(PMU)data as input to train deep neural networks to develop decision logics.Under normal conditions of the system,the distance relay will work as usual,but under pressure conditions,fault identification is required,and the distance relay will be operated only when it is identified as a fault.The proposed method is tested on a 9-bus system and compared with the traditional scheme.The test results show that this method can significantly reduce the maloperation of the distance relay and has a higher protection efficiency.

关 键 词：广域后备保护 深度学习 继电保护 

分 类 号：TM773[电气工程—电力系统及自动化]