基于Stacking集成学习的有源台区线损率评估方法  

A line loss rate evaluation method based on stacking ensemble learning for transformer district with DG

作　　者：董美娜 刘丽平[2] 王泽忠[1] 王守强 张子岩 邹运 Dong Meina;Liu Liping;Wang Zezhong;Wang Shouqiang;Zhang Ziyan;Zou Yun(School of Electrical and Electronic Engineering,North China Electric Power University,Beijing 102206,China;China Electric Power Research Institute,Beijing 100192,China)

机构地区：[1]华北电力大学电气与电子工程学院,北京102206 [2]中国电力科学研究院,北京100192

出　　处：《电测与仪表》2023年第6期134-139,173,共7页Electrical Measurement & Instrumentation

基　　金：国家电网有限公司科技项目(5600-201919168A-0-0-00)。

摘　　要：人工智能及机器学习的发展,为有源台区线损率的评估提供了崭新的思路。提出一种基于Stacking集成学习的有源台区线损率评估方法。从特定系统中提取有源台区数据,采用互信息等方法处理数据中异常值,并建立电气特征指标体系。考虑传统的机器学习与不同思想的集成学习算法之间的差异,综合线性模型与非线性模型,选择线性回归算法、随机森林算法、GBDT算法作为基学习器,构建多算法融合的Stacking集成学习模型。以某省有源台区数据为例,验证了所提方法的准确性和有效性。The development of artificial intelligence and machine learning provided a new idea for the evaluation of line loss rate of transformer district with DG.A line loss rate evaluation method based on Stacking ensemble learning for transformer district with DG was proposed in this paper.Data of transformer districts with DG was extracted from specific systems and the outliers in the data were processed by means of mutual information to establish the electrical characteristic indicator system,considering the difference between traditional machine learning and different ideas of ensemble learning algorithms,integrated linear model and nonlinear model,linear regression,random forest and GBDT were involved in base-learner layer,and the model based on multi-algorithm combination of Stacking ensemble learning was built,accuracy and effectiveness of the proposed method was confirmed based on the data of transformer districts with DG.

关 键 词：有源台区 线损率 互信息 集成学习 多算法融合 

分 类 号：TM764[电气工程—电力系统及自动化]