摘要
针对继电保护整定过度依赖人工的问题,提出基于深度学习的继电保护定值智能校验方法。首先,提出了有功无功潮流熵模型,辨识电网输电线路分布状态的关键程度;其次,给出了线路潮流转移的冲击度指标,度量电网安全稳定的边界,获得关键输电线路潮流转移导致的故障概率和输电线路潮流冲击保护能力;第三,将有功无功潮流熵、线路潮流转移冲击度以及电网拓扑、潮流状态、继电保护选择性、负荷等信息作为深度学习模型的输入,将故障概率和继电保护定值校验序列作为输出进行训练,获得不同场景下继电保护定值校验顺序模型。通过对IEEE57电网的仿真验证,表明了所提方法的有效性。
Aiming at the problem that relay protection setting depends too much on manual work,an intelligent verification method of relay protection setting based on deep learning is proposed.Firstly,an active and reactive power flow entropy model is proposed to identify the key degree of transmission line distribution state.Secondly,the impact index of line power flow transfer is given to measure the boundary of power grid security and stability,and the fault probability caused by key transmission line power flow transfer and transmission line power flow impact protection ability are obtained.Thirdly,the active and reactive power flow entropy,line power flow transfer impact,power grid topology,power flow state,relay protection selectivity,load and other information are taken as the input of the deep learning model,and the fault probability and relay protection setting verification sequence are taken as the output to train the relay protection setting verification model under different scenarios.The simulation of IEEE57 power grid shows the effectiveness of the proposed method.
作者
韩学军
HAN Xuejun(East China Branch of State Grid Corperatiom of China Ltd.,Shanghai 200120,China)
出处
《电子器件》
CAS
北大核心
2023年第5期1442-1448,共7页
Chinese Journal of Electron Devices
关键词
深度学习
潮流介数
继电保护
校验
智能
deep learning
power flow betweenness
relay protection
verification
intelligence
作者简介
韩学军(1964-),男,上海人,学士,高级工程师,研究方向:电力系统继电保护,hd_hxj@163.com。
