Abstract One of the most common problems in the electricity sector is the lack of accurate information about the structure of the low-voltage distribution network, particularly the association between the customers and the substation’s phases. Identifying to which substation’s phase each customer is connected reduces the response time to contingencies, improves the detection of technical and non-technical losses, and enables the application of load balancing techniques, among other benefits. This paper presents a new method for phase identification and substation detection based on a correlation analysis of the variations of load consumption. Our method achieves accurate results with much fewer samples than the previous works, which is effective even if there are a low percentage of smart meters installed and missing data. We use statistical tests to avoid getting erroneous results from non-significant correlations values. It was tested on a public dataset and validated using real measurements from a neighborhood in Tucuman, Argentina. In the case of 200 customers and one to four weeks of data, we obtained an average accuracy of 80–95% if only 50% of the customers are measured, or 93–98% if all the customers are measured.

中文翻译：

---

使用有限电力消耗测量数据分析进行相位识别和变电站检测

摘要 电力部门最常见的问题之一是缺乏关于低压配电网结构的准确信息，特别是用户与变电站相位之间的关联。识别每个客户连接到哪个变电站的阶段可以减少对突发事件的响应时间，改进技术和非技术损失的检测，并实现负载平衡技术的应用，以及其他好处。本文提出了一种基于负荷消耗变化相关性分析的相位识别和变电站检测的新方法。我们的方法以比以前的工作少得多的样本获得了准确的结果，即使安装的智能电表的百分比很低并且缺少数据，这也是有效的。我们使用统计测试来避免从不显着的相关值中得到错误的结果。它在公共数据集上进行了测试，并使用来自阿根廷 Tucuman 的一个社区的真实测量进行了验证。在 200 个客户和 1 到 4 周的数据的情况下，如果只测量了 50% 的客户，我们获得了 80-95% 的平均准确度，如果测量了所有客户，我们获得了 93-98% 的平均准确度。