摘要
近年来,随着分布式光伏的发展,用户可参与的户用光伏系统在低压配电网中的并网数量快速增长。针对高比例户用光伏下的低压配电网具备的基本特征,提出一种基于台区配变及用户功率平衡关系的光伏时序功率卷积模型,并在此基础上,采用基于高斯混合模型的电压聚类优化目标函数进行户变关系识别。首先,基于用户-配变功率平衡关系,建立了高比例户用光伏的低压户变关系的多元线性回归模型;其次,提出一种基于Bi LSTM算法的光伏时序识别方法,筛选出光伏瞬时渗透率最高的时序片段;根据已识别的光伏时序片段建立光伏权重矩阵,并引入该矩阵将多元线性回归模型进行加权回归后转化为光伏时序功率卷积模型求解,在平抑光伏波动影响的同时初步识别户变从属关系;然后,提出一种基于高斯混合模型进行电压聚类的优化目标函数作为验证步骤,能够进一步排除验证错误的用户,提高户变关系的准确率;最后,以某地区实际低压配电网为例,分别使用该地区推广户用光伏前后不同时期的历史数据集进行识别,通过识别结果对比,验证了所提方法的有效性和实用性。
In recent years,with the development of distributed PV,the number of household PV systems in which the users can participate has increased rapidly in the low-voltage distribution network(LVDN).Aiming at the essential characteristics of LVDN with a high proportion of household PV systems,in this paper,a time-sequential convolution model based on the power balance relationship between the transformers and the customers is proposed,and an optimized function of voltage clustering based on the Gaussian mixture model is used to identify the relationship.Firstly,based on the transformer-customer power balance relationship,a multiple linear regression model(MLR)of the transformer-customer relationship identification in the LVDN with a high proportion of household PV systems is established.Secondly,a PV time-sequential identification based on the BiLSTM algorithm is proposed to screen out the time sequences with the highest transient PV penetration.According to the identified PV time segments,the PV weight matrix is built which was introduced to transform the MLR model into the time-sequential convolution model by the weighted regression to identify the subordination between the transformers and the customers initially while stabilizing the impact of the PV fluctuations.Then as a verification step,an optimal objective function of voltage clustering based on the Gaussian mixture model(GMM)is proposed in order to eliminate the incorrect customers and improve the identification accuracy.Finally,taking an actual LVDN in a particular district as an example,the historical datasets in different periods before and after the extension of the household PV systems in this district are used for identification. The results are compared to verify the effectiveness and practicability.
作者
朱音洁
赵健
宣羿
孙智卿
ZHU Yinjie;ZHAO Jian;XUAN Yi;SUN Zhiqing(College of Electrical Engineering,Shanghai University of Electric Power,Yangpu District,Shanghai 200090,China;State Grid Hangzhou Power Supply Company,Hangzhou 310016,Zhejiang Province,China)
出处
《电网技术》
EI
CSCD
北大核心
2024年第3期1160-1171,共12页
Power System Technology
基金
国家重点研发计划项目(2020YFB1506804)
国家自然科学基金项目(51907114)。
关键词
低压配电网
户变关系
拓扑辨识
智能电表
户用光伏
low-voltage distribution network
transformer-customer relationship
topology identification
smart meter
household PV system
