摘要
针对光伏电站运行原始数据中异常数据占比高、数据总体质量差的特点,对数据的异常识别与清洗是进行数据分析、预测的前提。为此,分析了光伏电站辐射强度-功率异常数据的特征和来源,提出一种基于滑动标准差曲线线性拟合的变点检测法,以及一种变点-四分位联合的光伏功率异常数据识别算法。利用多个光伏电站数据验证了所提算法的有效性和普适性,实现了对零散型、堆积型等各类异常数据的良好检测。
In view of the characteristics of high proportion of abnormal data and poor overall quality of data in the original operation data of photovoltaic power station,the identification and cleaning of abnormal data is the premise of data analysis and prediction.Therefore,this paper analyzes the characteristics and sources of photovoltaic power station radiation intensity-power anomaly data,and proposes a change point detection method based on linear fitting of sliding standard deviation curve and a change point quartile joint photovoltaic power anomaly data recognition algorithm.The effectiveness and universality of the proposed algorithm are verified by the data of several photovoltaic power stations,and the good detection of scattered,stacked and other abnormal data is realized.
作者
马天东
耿天翔
李峰
钟海亮
MA Tiandong;GENG Tianxiang;LI Feng;ZHONG Hailiang(State Grid Ningxia Electric Power Co.,Ltd.,Yinchuan 750001,China;Electric Power Research Institute of State Grid Ningxia Electric Power Co.,Ltd.,Yinchuan 750001,China)
出处
《微型电脑应用》
2024年第6期105-108,共4页
Microcomputer Applications
基金
国家电网公司科技项目(5229NX20007Z)。
关键词
光伏电站
异常检测
光伏功率
变点检测
四分位
photovoltaic power station
anomaly detection
photovoltaic power
change point detection
quartile
