Open AccessImputation of Missing Values for Solar Irradiance Data under Different Weathers using Univariate Methods
Author(s) -
N. B. Mohamad,
Boon-Han Lim,
An-Chow Lai
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/721/1/012004
Subject(s) - missing data , imputation (statistics) , univariate , solar irradiance , irradiance , statistics , photovoltaic system , mathematics , interpolation (computer graphics) , meteorology , environmental science , computer science , geography , multivariate statistics , engineering , animation , physics , computer graphics (images) , quantum mechanics , electrical engineering
Significant investment risks of large photovoltaic (PV) systems are uncertainties of energy yield predictions from a PV power plant. Unfortunately, solar irradiance weather datasets often have missing values due to operational issues; hence techniques of imputing or recovering the missing values become essential. This paper examined five statistical imputation methods that are frequently adopted in missing values analysis for daily solar irradiance series based on two different weather conditions in the tropical climate with 10% to 50% of missing values and the methods were evaluated using performance indicators such as normalised root-mean-square at standard test conditions (nRMSE STC ). The results show that during Sunny weather, the Bezier curve and Stineman interpolation give the lowest nRMSE STC for 10% and 20% to 50% of daily missing values, respectively. Meanwhile, during Largely Cloudy weather, three methods share the best estimations, which are Linear interpolation for 10% missing values, Stineman interpolation for 20%, 30% and 50% missing values, and Spline interpolation for 40% of missing values.