摘要
新能源功率预测是提高新能源场站控制,保障高比率新能源发电接入电网安全稳定运行的关键技术。目前,由于通信故障、设备异常、人为限电等不确定性问题,导致新能源场站的实测数据中含有高比例异常数据,进而降低了功率预测的精度。有效的数据清洗可以提高数据质量,使新能源功率预测结果更加精确。本文首先概述了数据清洗的主流方法;然后对异常数据进行详细分类,从异常值剔除和缺失值重构两个方面重点阐述和分析了现有数据清洗方法的基本思路、应用条件以及优缺点;最后指出了未来数据清洗中值得关注的问题和方向。
Power prediction is an essential technique for the safe and stable use of renewable energy,which can improve the control performance when the penetration of renewable energy power generation in the system becomes significant.However,large amounts of abnormal data lead to a decrease in prediction accuracy.Through efficient data cleaning,the precision of power prediction can be improved.Therefore,this paper focuses on the review and prospect of data cleaning for the renewable energy power prediction.Firstly,this paper introduces the methods of data cleaning.After the abnormal data are classified,the basic ideas and the application conditions of data cleaning are elaborated and analyzed from two perspectives,which are the rejection of the abnormal values and reconstruction of the missing values.Finally,the key problems in data cleaning for renewable energy are proposed.
作者
武佳卉
邵振国
杨少华
肖颂勇
吴国昌
Wu Jiahui;Shao Zhenguo;Yang Shaohua;Xiao Songyong;Wu Guochang(College of Electrical Engineering and Automation,Fuzhou University,Fuzhou 350108;Fujian Smart Electrical Engineering Technology Research Center,Fuzhou 350108;State Grid Putian Electric Power Supply Company,Putian,Fujian 351100;Putian Li Yuan Group Company,Putian,Fujian 351100)
出处
《电气技术》
2020年第11期1-6,共6页
Electrical Engineering
基金
国家自然科学基金(51777035)
福建省自然科学基金(2016J01019)。
关键词
新能源功率预测
不确定性
数据清洗
异常值剔除
缺失值重构
renewable energy power prediction
uncertainty
data cleaning
rejection of the abnormal values
reconstruction of the missing values
