浙江电力

2020, v.39;No.286(02) 83-89

[打印本页] [关闭]
本期目录(Current Issue) | 过刊浏览(Archive) | 高级检索(Advanced Search)

基于HCM的光伏反窃电算法研究
Research on HC-based PV Power Anti-theft Algorithm

陈海峰,应国德,曹杰,林超,潘成峰,金潮
CHEN Haifeng,YING Guode,CAO Jie,LIN Chao,PAN Chengfeng,JIN Chao

摘要(Abstract):

针对光伏窃电行为,设计了光伏反窃电算法。该算法首先将发电曲线分段处理,利用傅里叶级数对曲线进行初步处理;然后提取出时间、方差、增量等7项与数值大小本身无关的量,以及3项特定的发电量作为曲线特征,设计了聚类中心的选择方法,制订了正常、故障、窃电等3种聚类中心的参数;再根据各特征分量的生成原理,依次进行HCM(硬聚类);最后聚类至窃电类的样本,就是疑似窃电的用户。此外,在能够获取气象、配电网、设备基础信息的场景下,设计了更细粒度的分析算法。所设计算法已在浙江台州得到实际应用,并取得良好效果。
A photovoltaic power anti-theft detection algorithm is designed to fight against PV power theft. The algorithm segments the generation curve and preliminarily process it with Fourier series. Then, seven items,such as time, variance and increment that are unrelated to the value itself and three specific amounts of generation are extracted as the characteristics of the curve to design the selection method of the clustering center and establish parameters of three clustering centers such as normal, fault and theft. Then, according to the generation principle of each characteristic component, HCM(hard C-means) clustering is carried out successively; finally, the clustered theft samples are the theft suspects. Besides, a more fine-grained analysis algorithm is designed when the weather, distribution network and equipment can be obtained. The algorithm was already applied in Taizhou, and good results are obtained.

关键词(KeyWords): 光伏;窃电;傅里叶级数;特征;硬聚类
PV;power theft;fourier series;characteristic;HCM

Abstract:

Keywords:

基金项目(Foundation):

作者(Author): 陈海峰,应国德,曹杰,林超,潘成峰,金潮
CHEN Haifeng,YING Guode,CAO Jie,LIN Chao,PAN Chengfeng,JIN Chao

DOI: 10.19585/j.zjdl.202002013

参考文献(References):

扩展功能
本文信息
服务与反馈
本文关键词相关文章
本文作者相关文章
中国知网
分享