基于谱聚类的电压波动图谱一致性线变关系检测分析应用

Application of Detection and Analysis on The Relationship of Circuit and Transformer of Voltage Fluctuation Spectrum Based on Spectral Clustering

基于谱聚类的电压波动图谱一致性线变关系检测分析应用是在电力物联网背景下,充分挖掘配网台区运行数据价值,通过大数据分析技术进行的探索和实践。根据"同一馈线下台区之间电压波动变化趋势一致"的原理,借助谱聚类分析技术,将月度1000万余条的电压运行数据,转换为"电压波动指纹图谱",构建基于谱聚类的电压波动图谱一致性线变关系检测模型,对同一馈线下台区电压时序数据进行波动一致性识别,输出馈线下电压波动明显异于其他台区的异动台区清单,为错误档案的系统化甄别提供快速、准确的分析结论,从大数据视角为检测线变异常关系提供全新手段。一方面切实有效的指导开展线变关系治理,进一步促进降低线损、落实基层减负;另一方面可为客户可靠供电和精准营销提供基础保障、改善客户体验;也为感知层和应用层提供坚实的数据质量监管提供手段。目前完成模型构建并持续优化,并现场试点验证,准确核实出27个线变拓扑关系问题。

Analysis and application on detecting the consistency of relationship between circuit and transformer based on spectral clustering,which is explored and practiced under the Background of Building Ubiquitous Power Internet by Gansu Company,fully excavate the value of distribution network operation data with the help of the large data analysis technology.According to the principle of"the trend of voltage fluctuation is the same between the substations under the same feeder",With the help of spectral clustering analysis technology,more than 10 million monthly voltage operation data are transformed into"voltage fluctuation fingerprints"Constructing a consistent line-variation relationship detection model based on spectral clustering for voltage fluctuation maps,and identifying the fluctuation consistency of voltage timing data of the same feeder under the station,output list of dislocation station,which voltage fluctuation under feeder is obviously different from other stations,providing fast and accurate analysis conclusions for the systematic identification of false files,and providing a new means for detecting abnormal relationship between lines from the perspective of big data.On the one hand,it effectively guides the development of line-change relationship management,further promotes the reduction of line losses and the implementation of grassroots burden reduction;On the other hand,it can provide basic guarantee for customers’reliable power supply and precise marketing,and improve customer experience;For another,the result provides reliable ways of data quality regulation for perception level and application level.At present,the model is built and continuously optimized.The on-site pilot verification has been carried out in the city and state companies,and 27 line-topological relationships have been accurately verified.relationships have been accurately verified.