基于密度聚类的低压台区归属关系及相位识别方法  

Identification of Low-Voltage Distribution Network Attribution Relationship and Phase Information Based on Density Clustering

作　　者：闫东辉 YAN Donghui(Hainan Power Grid Co.,Ltd.,Sanya 572000,Hainan,China)

机构地区：[1]海南电网有限责任公司,海南三亚572000

出　　处：《南方能源建设》2023年第5期149-156,共8页Southern Energy Construction

摘　　要：[目的]供电部门记录的正确的拓扑信息有助于工作人员监测电网信息,分析故障,优化电网运行以满足低压配电台区精益化、智能化管理的需要。目前,各式新型用电设备及用户的加入使低压配电网络结构呈现出持续变化的特征,线路维护成本被大大提高。[方法]为此,提出基于密度聚类的低压台区归属关系识别方法。首先,提取智能电表有效电压数据生成高维时序电压矩阵;其次,采用t分布随机近邻嵌入方法(t-distributed Stochastic Neighbor Embedding,t-SNE)对高维时序电压数据进行特征提取与降维;然后,应用基于数据密度的噪声应用空间聚类方法(Density-Based Spatial Clustering of Applications with Noise,DBSCAN)对降维后的数据进行聚类分析,实现低压用户台区归属信息的识别;最后,对海南省三亚市某台区实际数据进行分析,并将所提方法与其他主流的拓扑识别算法进行对比。[结果]分析结果表明所提方法能够达到95%以上的台区识别准确率,高于目前其他主流的拓扑信息识别方法。[结论]文章中的方法在解决此类问题上具有有效性与优势性,可以为实际工程应用提供参考,为低压台区拓扑信息识别领域提供不一样的研究思路。[Introduction]The correct topology information recorded by the power supply department can help the staff monitor the power grid information,analyze the faults,and optimize the operation of the power grid to meet the needs of lean and intelligent management of low-voltage distribution networks.At present,the addition of various new types of electricity-using equipment and users has caused the low-voltage distribution network structure to show a continuous change in characteristics,and the line maintenance cost is greatly increased.[Method]Therefore,the identification method of low-voltage distribution network attribution relationship based on density clustering was proposed.First,the effective voltage data collected by smart meters were extracted to generate a high-dimensional time-series voltage matrix.Then,the t-distributed Stochastic Neighbor Embedding algorithm(t-SNE)and Density-Based Spatial Clustering of Applications with Noise algorithm(DBSCAN)were applied to cluster the voltage data to achieve identification of lowvoltage distribution network attribution relationship.Finally,the actual data of a low-voltage distribution network in Sanya City,Hainan Province were analyzed,and the proposed method is compared with other mainstream topology identification methods.[Result]The analysis results show that the proposed method can achieve more than 95%of identification accuracy,which is higher than other mainstream topology identification methods.[Conclusion]The proposed method is effective and advantageous in solving such problems,and can provide reference for practical engineering applications and offer a different research idea in the field of topology identification of low-voltage distribution network.

关 键 词：低压台区 电压数据信息 t分布随机近邻嵌入方法 基于数据密度的噪声应用空间聚类方法 台区归属关系识别 相位识别 

分 类 号：TM7[电气工程—电力系统及自动化] TM727