蔡永智 唐捷 危阜胜 李健 郭文翀 董志奎 杨银

摘要：低压台区拓扑关系识别是实现电网智能化的一个关键基础，相对于人工识别，从数据挖掘角度进行识别，具有成本低、准确率高、时效性好的优势。就其相序识别问题，提出了基于随机森林算法的低压台区相序关系自动识别方法。首先基于随机算法原理搭建了用于台区相序识别的随机森林算法模型，并提出采用 F1_score统计值作为识别模型的评价指标;然后将搭建好的模型应用到实际台区进行训练，训练方式设置了两类测试集：口袋内台区测试集与口袋外台区测试集;最后将训练好的算法模型对测试样本进行相序分许。实证结果证明，所提方法对台区的相序拓扑分析有較高准确性，为低压台区拓扑分析提供了一种技术思路和可行方法。

关键词：低压台区;相序识别;随机森林

中图分类号：TM71             文献标志码：A

文章编号：1009-9492（2021）12-0050-04

开放科学（资源服务）标识码（OSID）：

Phase Sequence Recognition Method of Station Area Based on Random Forest Algorithm

Cai Yongzhi1，Tang Jie2，Wei Fusheng2，Li Jian1，Guo Wenchong1，Dong Zhikui3，Yang Yin3

（1. Metrology Center of Guangdong Power Grid Co.， Ltd.， Guangzhou 510060， China;

2. Guangdong Power Grid Corp， Guangzhou 510060， China;3. Guangzhou Power Electrical Technology Co.， Ltd.， Guangzhou 510700， China）

Abstract： The recognition of low-voltage station topology is a key foundation for the realization of intelligent power grid. Compared with manual recognition， recognition from the perspective of data mining has the advantages of low cost， high accuracy， and good timeliness. Regarding the phase sequence identification problem， an automatic identification method of the phase sequence relationship of low-voltage stations based on the random forest algorithm was proposed. First， based on the principle of random algorithm， a random forest algorithm model for phase sequence recognition of station area was built， and the F1_score statistical value was proposed as the evaluation index of the recognition model. Then the built model was applied to the actual station area for training. The training method was set with two types of test sets： the pocket inner station area test set and the pocket outer station area test set. Finally， the trained algorithm model was divided into the phase sequence of the test samples. The empirical results prove that the proposed method has high accuracy in the phase sequence topology analysis of the station area， then provides a technical idea and feasible method for the topology analysis of the low voltage station area.       Key words： low-voltage station area; phase sequence recognition; random forest

0 引言

随着社会的发展，用户对更可靠更优质的电能供应、更优质更贴心的供电服务的要求日益提升[1-2]。而技术进步推进了智能配电网建设快速发展，给电网公司管理电网运行及提升服务提供了技术支撑。配电网的拓扑结构是配电网进行线损分析、故障诊断、潮流计算、三相平衡等应用功能的基础。随着智能电表的普及，终端用户用电数据的积累，这使得能够对台区数据进行挖掘，研究台区运行数据高频采集、户变、相序与线户关系识别算法研究变为可能。……