摘要
目前含新能源电网短路电流超标问题日益严重,因其状态变化更快、幅度更大,离线的短路电流超标分析可能无法穷尽超标场景,在线实时分析具有相当的必要性。考虑到主流物理模型计算方法难以满足在线计算速度需求,更快速的计算具有重大意义,因此,提出一种数据—物理融合驱动的短路电流超标评估策略。首先,在分析影响短路电流主要因素的基础上,为了提高计算速度,将原先特征降维为仅考虑负荷的影响,再结合最优潮流及随机模拟生成大量样本集合,通过机器学习算法训练得出数据驱动模型,在此基础上,使用漏检率和误检率对数据模型进行阈值整定;然后,利用数据驱动模型初步筛选短路电流超标场景;最后,通过最新研究中所提理论物理模型对初筛后的短路电流场景进行高精度校验,并在含光伏电源的IEEE 39节点模型上进行验证。仿真结果表明:该策略可在确保不遗漏超标短路电流场景的前提下有效提升校验速度。
The problem of over-limit short-circuit current in power grid containing renewable energy is becoming increasingly serious.Because of its faster and larger state changes,offline over-limit short-circuit current analysis may not be able to cover all the over-limit scenarios.Therefore,online analysis is quite necessary.Considering that the mainstream physical model calculation method can hardly meet the online calculation speed demand,faster calculation is of great significance.Therefore,this paper proposes a strategy of combining data-driven and model-driven method for over-limit short-circuit current evaluation for power grid with high penetration of renewable energy.Firstly,based on the analysis of the main factors affecting the short-circuit current,in order to improve the calculation speed,the original dimension is reduced to consider only the influence of load.Then,the optimal power flow and random simulation methods are combined to generate a large set of samples,and the data-driven model is obtained through machine learning algorithm training.On this basis,the error analysis and threshold setting of the model are carried out by using false positive rate and false negative rate.Then,the data-driven model is used to screen over-limit short-circuit current scenarios;Finally,the theoretical physical model proposed in the latest research is used to verify the short-circuit current scenario after preliminary screening with high accuracy.It is verified on the IEEE 39 bus model with photovoltaic power supply.The simulation results show that this strategy can effectively improve the verification speed without omitting the over-limit short-circuit current scenarios.
作者
熊志
章谋成
姚伟
乔立
赵红生
刘巨
王博
XIONG Zhi;ZHANG Moucheng;YAO Wei;QIAO Li;ZHAO Hongsheng;LIU Ju;WANG Bo(Economic and Technical Research Institute,State Grid Hubei Electric Power Co.,Ltd.,Wuhan 430077,China;College of Electrical Engineering&New Energy,China Three Gorges University,Yichang 443002,China;State Key Laboratory of Advanced Electromagnetic Engineering and Technology,Huazhong University of Science and Technology,Wuhan 430074,China)
出处
《电力科学与技术学报》
CAS
CSCD
北大核心
2023年第4期24-34,共11页
Journal of Electric Power Science And Technology
基金
国家自然科学基金(52107095)。
关键词
新能源
数据—物理融合
机器学习
数据驱动
短路电流超标
renewable energy
data-physical fusion
machine learning
data-driven
over-limit short-circuit current