电力系统及其自动化学报2017，Vol.29Issue(9)：27-33,7.DOI:10.3969/j.issn.1003-8930.2017.09.005
计及用电模式的居民负荷梯度提升树分类识别方法
Gradient Boosting Decision Tree Method for Residential Load Classification Considering Typical Power Consumption Modes
摘要
Abstract
Classification and identification for residential load are the basis of load monitoring and demand-side manage?ment. In order to realize the extraction and identification of the load power consumption modes for residents ,a gradient boosting decision tree method for load classification and identification,which takes typical power consumption modes into account,is proposed in this paper by using principal component analysis(PCA)to deduce and aggregate the com?mon dataset of load. Firstly,the load data are resampled and the characteristic samples of various types of load energy consumption data are obtained,which are further normalized and reduced by PCA to acquire the principal components. Secondly,by means of improved K-means clustering method,typical power consumption modes are obtained. More?over,gradient boosting decision tree is trained with super-parameter optimization,and the test set is used to identify dif?ferent types of loads. Through the tests on the common dataset and measured data,it is found that the proposed method has good effect on load classification and identification.关键词
公共数据集/负荷用电模式/改进K均值聚类/梯度提升树Key words
public dataset/power consumption mode/improved K-means clustering/gradient boosting decision tree分类
信息技术与安全科学引用本文复制引用
王守相,刘天宇..计及用电模式的居民负荷梯度提升树分类识别方法[J].电力系统及其自动化学报,2017,29(9):27-33,7.基金项目
天津市应用基础与前沿技术研究计划资助项目(14JCYBJC21100) （14JCYBJC21100）
