High-Precision Forecasting Model of Solar Irradiance Based on Grid Point Value Data Analysis for an Efficient Photovoltaic System

Abstract

An accurate forecasting system is extremely crucial in order to simulate an optimum output level of photovoltaic (PV) power production for the next day. In this study, a relatively high-precision model of solar irradiance forecasting based on grid point value (GPV) datasets using relative humidity, precipitation, and three-level cloud covers parameterization has been conducted in Hitachi and four main cities in Japan. In the case of cloudy/rainy/snowy days, the influence of liquid water path is further introduced to the model. As a result, correlation coefficient r of 0.94, 0.91, 0.91, 0.89, and 0.92 have been obtained using 21UTC forecast version in 2012 datasets for Hitachi, Tokyo, Nagoya, Osaka, and Fukuoka, respectively. Surprisingly, although the earlier forecast version, using 9UTC datasets, was later applied to the model, there was no significant change to the r for these five locations as their values reduced by only approximately 0.01 at most. Furthermore, a similar trend has also been observed for the 2013 datasets from a comparison of 21UTC and 9UTC versions, which highly supports the fact that this model is reliable, since it still remains in a high-precision state even in the case where the earlier datasets of previous day are used.