基于图滤波器和SVDD算法的分布式光伏集群发电异常检测研究

Research on Anomaly Detection of Distributed Photovoltaic Cluster Power Generation Based on Graph Filtering and SVDD Algorithm

由于电网企业不断加快数字化转型,利用北斗定位技术将自动获取区域内光伏计量装置经纬度这一关键技术参数。文章充分利用分布式光伏集群内光伏发电装机位置空间相关性,提出一种在弱监督下基于图滤波与支持向量数据描述(support vector data description,SVDD)的分布式光伏集群发电异常检测方法。首先建立分布式光伏集群发电图数据结构模型,通过加权邻接矩阵描述分布式光伏发电点空间耦合性,其次构造图高通滤波器将时域参数转化为频域参数,然后通过SVDD算法优化图滤波结果,进一步挖掘图高通滤波器阈值与输出功率数据之间的关系。结果表明,采用图滤波器和SVDD算法模型方法在分布式光伏发电异常检测精度上有显著提高。

Due to the continuous acceleration of digital transformation by power grid enterprises,Beidou positioning technology is used to automatically obtain the key technical parameter of longitude and latitude of photovoltaic metering devices in the region.A distributed photovoltaic cluster power generation anomaly detection method based on graph filtering and support vector data description(SVDD)is proposed with weakly supervised algorithm,which fully taking advantage of the spatial correlation of photovoltaic power generation installation locations within distributed photovoltaic clusters.Firstly,a graph data structure of distributed photovoltaic cluster power generation is established to describe the spatial coupling of distributed photovoltaic power generation points through adjacency matrix.Secondly,a graph high-pass filter is constructed in order to achieve preliminary detection of photovoltaic power generation anomalies.Thirdly,the SVDD algorithm is applied to optimize the graph filtering results,further exploring the relationship between the threshold of the graph high-pass filter and the output power data.Finally,the simulation is carried out with actual data and the results indicate that graph filtering and SVDD model methods can significantly improve the accuracy of anomaly detection in distributed photovoltaic power generation.