摘要
为了对光伏电站的历史数据进行修复,文中建立了基于三次样条插值和BP神经网络的光伏电站出力数据修复模型。首先对光伏电站原始数据进行清洗和处理;其次,寻找一天中只有一个数据空白点的情况,采用光滑样条插值法对其进行插补,然后寻找一天中有多个数据空白点的情况,采用相似日绝对均值法对其进行插补;再次,根据出力数据的特征要素及BP神经网络模型对待插补数据进行预测插补;最后,输出修复后的光伏电站出力数据。以青海海西地区光伏电站为例进行仿真验证,证明了模型的可行性及正确性。对光伏电站的历史数据进行修复,保证了数据的完整性和真实性,为研究光伏发电相关问题奠定了基础。
A photovoltaic power generation output data repair module is built based on cubic spline interpolation and BP neural network for the purpose of repairing historical date of a photovoltaic plant.The first step is the photovoltaic power station raw data cleaning and processing.Secondly,find a daily data with one data blank point and interpolate data by smooth spline interpolation method then find the daily data with data blank points and interpolate data by similar daily absolute mean method.Thirdlyfpredictive data interpolation is adopted based on the characteristic elements of output power and BP neural network.Finallyfoutput teh repaired data of photovoltaic power station.Photovoltaic power stations in Haixi Qinghai were taken as examples to validate the feasibility and correctness of the model.The repaired photovoltaic power generation output data ensured its completeness and authenticityfand lay foundation for its related topics.
作者
赵洪斌
ZHAO Hongbin(Qinghai Institute of Water Resources and Hydropower Survey and Design,Xining 810001,China)
出处
《青海大学学报》
2020年第1期68-74,共7页
Journal of Qinghai University
关键词
光伏出力
三次样条插值
BP神经网络
相似日绝对均值
数据修复
photovoltaic power output
cubic spline interpolation
BP neural network
similar daily absolute mean method
data repair
