摘要
电力企业对发电情况的研究一般基于发电统计数据,但统计数据中的异常数据会影响对发电情况的正确认知。针对发电统计数据的异常辨识难点,剖析了统计数据的组成结构及相互关系,分别依据自身历史数据或同类电厂数据提出了传统电源和新能源的横向与纵向聚类方法,确立了不同类型电源月度发电统计数据的异常辨识条件,提出了面向不同电源特征的发电统计数据异常辨识方法。通过实例分析,验证了该方法的合理性与有效性。
Researches on power generation of power enterprises are generally based on the statistical data. However,the correct perception of the power generation will be affected by the abnormal data in the statistical data. In this study, a data abnormality identification method for statistical data based on different power characteristics was proposed. The structures and relationships of statistical data have been analyzed aiming at identifying the abnormal data from the statistical data. To establish the anomaly identification conditions of monthly generating statistical data of different types of power sources, the horizontal and vertical clustering methods of traditional power and new energy sources have been proposed based on their own historical data or similar power plant data, respectively. The rationality and effectiveness of the method have been verified by example analysis.
作者
高骞
赵阿南
汤奕
GAO Qian;ZHAO Anan;TANG Yi(State Grid Jiangsu Electric Power Co.,Ltd.Nanjing 210029,China;School of Cyber Science&Engineering,Southeast University,Nanjing 210096,China;School of Electrical Engineering,Southeast University,Nanjing 210096,China)
出处
《供用电》
2020年第1期73-79,共7页
Distribution & Utilization
基金
国家自然科学基金项目（51577030）
国网江苏省电力有限公司管理咨询项目“国网江苏电力发电统计数据分析与质量管控体系研究咨询项目”~~
关键词
异常辨识
特征聚类
平稳性检验
三点平滑原理
电源特征
data abnormality identification
feature clustering
stationarity test
smoothing principle with three-point
characteristics of generation
