直流馈入下新能源高占比受端电网暂态电压稳定评估  

Transient voltage stability assessment of high proportion receiving end power grid of new energy under DC input

作　　者：崔晓丹 吴家龙 王彦品 许剑冰 冯佳期 Cui Xiaodan;Wu Jialong;Wang Yanpin;Xu Jianbing;Feng Jiaqi(NARI Group(State Grid Electric Power Research Institute)Co.,Ltd.,Nanjing 211106,China;National Key Laboratory of Power Grid Operation Risk Defense Technology and Equipment,Nanjing 211106,China)

机构地区：[1]南瑞集团(国网电力科学研究院)有限公司,江苏南京211106 [2]电网运行风险防御技术与装备全国重点实验室,江苏南京211106

出　　处：《能源与环保》2025年第2期191-202,共12页CHINA ENERGY AND ENVIRONMENTAL PROTECTION

基　　金：国家电网有限公司科技项目(4000-202224069A-1-1-ZN)。

摘　　要：实际运行的电力系统获取的量测数据大多是稳定的,失稳情况少。但是基于数据驱动的暂态稳定评估准确度与稳定和失稳有效样本量的平衡度有密切关系。为了克服评估模型对稳定样本的倾向性,将稳定和失稳样本的有效样本数量这一特征引入交叉熵损失函数,用来度量损失函数中不同类别的权重,训练双向长短期记忆网络模型。此外,基于文中提出的模型建立了分类判定阀值与评估结果的映射关系。当系统的运行工况发生变化时,采用迁移学习方法并将核主成分分析法应用于源域、目标域输入特征的提取,解决异构迁移学习的问题,仅需目标域的少量样本对源域模型进行微调,使模型能够快速适应变化场景下的暂态电压稳定评估。最后,在引入直流和新能源发电的IEEE39节点系统上进行实验,验证了所提方法的有效性。The measurement data obtained from the actual operation of the power system is mostly stable,with few cases of instability.However,the accuracy of data-driven transient stability assessment is closely related to the balance between the effective sample size of stability and instability.In order to overcome the tendency of the assessment model to stable samples,the feature of effective sample size of stability and instability was introduced into the cross-entropy loss function,which is used to measure the weights of different classes in the loss function,and to train the bidirectional long short-term memory(Bi-LSTM)network model.Moreover,the mapping relationship between the classification decision threshold and the assessment results was established based on the proposed model.When the operating conditions of the system changed,the transfer learning method was adopted and the kernel principal component analysis was applied to the extraction of input features in the source and target domains to solve the problem of heterogeneous transfer learning,and only needing a small number of samples from the target domain to fine-tune the source domain model,so that the model could be adapted quickly to the transient voltage stability assessment under the changing scenarios.Finally,experiments were conducted on the IEEE39 node system introducing DC transmission and new energy generation to verify the effectiveness of the proposed method.

关 键 词：暂态电压稳定评估 双向长短期记忆网络 样本不平衡 类别平衡损失函数 迁移学习 

分 类 号：TM712[电气工程—电力系统及自动化]