Open AccessRecovery of the Solar Irradiance Data using Artificial Neural Network
Author(s) -
Kah-Ching Ho,
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 - 10.1088/1755-1315/721/1/012006
Subject(s) - artificial neural network , photovoltaic system , renewable energy , irradiance , solar irradiance , computer science , solar energy , environmental science , reliability engineering , meteorology , engineering , electrical engineering , artificial intelligence , physics , quantum mechanics
Global solar irradiance (GHI) data plays a major role in the design, performance assessment, and monitoring of a solar energy conversion system. However, data loss happens sometimes due to various reasons. Hence, it is important to recover the lost data. Unfortunately, due to the high cost of measurement devices, other meteorological parameters are often not available to aid the data recovery. Nevertheless, a photovoltaic (PV) system could be installed nearby the site which the output power is recorded. This paper explored the capability of using output power from a PV panel to predict the lost or unavailable GHI data in the past via the Artificial Neural Network (ANN) technique. 1 day, 10 days, and 30 days of GHI data, time, and output power of a solar panel in July 2020 were used to train three ANN models. The ANN models were then used to predict one day and thirty days GHI data in July 2019 based on the PV output power of that period. The result shows that the ANN model with a higher number of training data can predict the GHI data with lower statistical errors. The results prove that PV output power can be used to recover the unavailable GHI data in the past with acceptable accuracy. This technique is useful for the industry to recover the historical time series of GHI for designing a new renewable energy project or assessing the past performance of a renewable energy project.