Photovoltaic forecasting model in Thailand: case study solar farm at Nakhon Ratchasima province

Abstract

The photovoltaic module performance was normally tested at the standard conditions. The use of the actual field performance has not been investigated. One purpose of this study was to find the predicted equation for forecasting the energy produced from photovoltaic modules. Data for this study were collected using the data from the solar farm in Nakhon Ratchasima province: solar irradiation, ambient temperature, module temperature and wind velocity. The daily composite data were 1,461 from 1st January 2015 – 31st December 2018. The five predicted equations, including both the main effect and the interaction effect, were statistically analysed by regression method using the Minitab software program for multiple linear regression. The requiring dependent variable (y) was the energy produced from the photovoltaic module. The four independent variables were solar irradiation (x1), ambient temperature (x2), module temperature (x3) and wind velocity (x4). For prediction preciscion, all the predicted equations were validated with the new gathered data from 1st January 2019 – 30th June 2019. The analysis results showed, the forecasting model together with the appropriate predicted equation with the highest coefficient of determination (R2 of 0.9873) and the standard deviation of prediction (S = 2.67%). This predicted equation, ŷ = 5258 + 5310.0 x 1 - 100.31 x 2 + 66.2x 4, consisted of the three external independent variables; solar irradiation, ambient temperature, and wind velocity. The independent variable sensitivities were also determined. The solar irradiation was the most sensitive to the predicted equation. Moreover, this predicted equation would be suggested to utilize as the estimator for the energy produced from the new photovoltaic power plant to be installed at the north-eastern part of Thailand.