| 引用本文: | 杨明,毛一风,李鹏,欧朱建.基于欠完备电压数据的户变关系识别方法[J].电力自动化设备,2024,44(12):100-107. |
| YANG Ming,MAO Yifeng,LI Peng,OU Zhujian.Transformer-customer relationship identification method based on incomplete voltage data[J].Electric Power Automation Equipment,2024,44(12):100-107. |
|
| 摘要: |
| 当前采集的台区用户数据存在一定比例的缺失，使得传统基于数据驱动的户变关系核查方法难以实际应用。为此，提出一种基于欠完备电压数据的户变关系识别方法。通过掩码自编码器直接从欠完备电压数据中提取特征编码；对特征编码应用参数自适应的基于密度的噪声应用空间聚类(DBSCAN)算法进行聚类，实现低压台区户变关系的准确识别。所提方法无须进行缺失数据填充，避免了数据填充误差对特征提取效果的影响。同时，该方法无须人工调参，降低了工程应用难度，提升了低压台区管理自动化水平。算例结果表明，在缺失率达80 % 的情况下，所提方法的户变关系识别准确率依然可达80 %。 |
| 关键词: 低压台区 户变关系识别 缺失数据 掩码自编码器 DBSCAN 参数自适应 |
| DOI：10.16081/j.epae.202409022 |
| 分类号:TM73 |
| 基金项目:国家自然科学基金资助项目(52207118)；山东省自然科学基金资助项目(ZR2022QE145)；泰山学者工程专项经费资助项目 |
|
| Transformer-customer relationship identification method based on incomplete voltage data |
|
YANG Ming1, MAO Yifeng2, LI Peng1, OU Zhujian3
|
|
1.Key Laboratory of Power System Intelligent Dispatch and Control of Ministry of Education, Shandong University, Jinan 250061, China;2.Changzhou Power Supply Company, State Grid Jiangsu Electric Power Co.,Ltd.,Changzhou 213000, China;3.Nantong Power Supply Company, State Grid Jiangsu Electric Power Co.,Ltd.,Nantong 226000, China
|
| Abstract: |
| There is a certain proportion of missing data in the currently collected customer data in low voltage(LV) station area, which makes it difficult to apply traditional data-driven methods to identify the transformer-customer relationship in practice. Aiming at this problem, a transformer-customer relationship identification method based on incomplete voltage data is proposed. Feature vectors are extracted directly from the incomplete voltage data by masked autoencoder. The density-based spatial clustering of applications with noise(DBSCAN) algorithm with self-adaptive parameters is used to cluster the feature vectors, thereby achieving accurate identification of transformer-customer relationship in LV station area. The proposed method does not need to perform missing data imputation, thereby avoiding the impact of data reconstruction error on feature extraction results. Additionally, this method requires no manual parameter adjustment, reducing the difficulty of engineering application and improving the automation management level of LV station area. The case study results show that the transformer-customer relationship identification accuracy of the proposed method can still achieve 80 % when the missing rate reaches 80 %. |
| Key words: low voltage station area transformer-customer relationship identification missing data masked autoencoder DBSCAN parameter self-adaption |
