东北电力技术2026，Vol.47Issue(2)：18-26,9.
基于DeepSMOTE和深度聚类的专变用户负荷辨识方法研究
Research on Load Identification Method of Special Transformer Users Based on DeepSMOTE and Deep Clustering
摘要
Abstract
According to the problem of class imbalance in the load data of specialized variable users and the limitations of traditional clustering methods in dealing with high-dimensional time series data,it proposes a method for identifying unbalanced load data of spe-cialized variable users based on deep oversampling(DeepSMOTE)and deep clustering to enhance the ability of special transformer users load identification.Firstly,it proposes a load data expansion method based on DeepSMOTE to achieve the balance of the dedica-ted transformer user electricity consumption data set.Then,it uses the deep clustering algorithm to cluster the load curve of the special transformer user,and extracts the typical load curve of the special transformer user.Finally,the experimental results show that the proposed method can realize the accurate identification of unbalanced load data sets.The identification method for imbalanced load data of specialized variable users based on DeepSMOTE and deep clustering has an accuracy rate as high as 93.2%,which can effectively improve the identification accuracy of imbalanced load samples.关键词
类别不平衡/负荷辨识/深度聚类/专变用户Key words
class imbalance/load identification/deep clustering/special transformer user分类
管理科学引用本文复制引用
吴林桥,荆澜涛,李金阔,田瑞,王亮..基于DeepSMOTE和深度聚类的专变用户负荷辨识方法研究[J].东北电力技术,2026,47(2):18-26,9.基金项目
创新能力提升联合基金(2022-NLTS-16-05) （2022-NLTS-16-05）
辽宁省教育厅基本科研项目(LJKMZ20221707) （LJKMZ20221707）
