基于K-Medoids聚类的分布式光伏台区线损异常感知算法

为保证分布式光伏台区稳定运行，精准有效地划分台区线损数据，提出基于K-Medoids聚类的分布式光伏线损异常感知算法，精准判断分布式台区线损异常程度。采用局部异常因子（LOF）算法判断分布式光伏台区数据局部异常程度，并筛选和去除受孤立点影响产生的异常线损数据。采取K-Medoids聚类方法聚类分析筛选后的分布式光伏台区数据，将异常线损率区间结合异常线损数据的聚类中心和欧式距离，完成台区线损异常感知。并创新性地引入粒度计算优化K-Medoids聚类算法聚类中心，提升异常数据感知效果。试验结果表明，所提算法可有效避免孤立点对异常感知效果的影响，精准有效地感知分布式光伏台区线损异常，并清晰划分台区线损数据类别。

Distributed Photovoltaic Station Area Line Loss Anomaly Sensing Algorithm Based on K-Medoids Clustering

In order to ensure the stable operation of the distributed photovoltaic station area and accurately and effectively divide the line loss data of the station area, a distributed photovoltaic line loss anomaly sensing algorithm based on K-Medoids clustering is proposed to accurately judge the degree of line loss anomaly in the distributed station area. The local anomaly factor (LOF) algorithm is used to judge the degree of local anomaly in the distributed photovoltaic station area data, and the anomalous line loss data generated by the influence of isolated points is filtered and removed. The distributed photovoltaic station area data after the filtering is clustered and analyzed by the K-Medoids clustering method. The anomalous line loss rate interval is combined with the clustering center and Euclidean distance of the anomalous line loss data, and the line loss anomaly sensing of the station area is completed, and the granularity calculation is innovatively introduced to optimize the K-Medoids clustering algorithm clustering center and to improve the sensing of anomalous data. The test results show that the proposed algorithm can effectively avoid the influence of isolated points on the anomalous sensing effect, accurately and effectively perceive the line loss anomalies in the distributed photovoltaic station area, and clearly divide the line loss data categories in the station area.