Solar Photovoltaic Power Prediction Using Big Data Tools

Abstract

Solar photovoltaic (PV) installation has been continually growing to be utilized in a grid-connected or stand-alone network. However, since the generation of solar PV power is highly variable because of different factors, its accurate forecasting is critical for a reliable integration to the grid and for supplying the load in a stand-alone network. This paper presents a prediction model for calculating solar PV power based on historical data, such as solar PV data, solar irradiance, and weather data, which are stored, managed, and processed using big data tools. The considered variables in calculating the solar PV power include solar irradiance, efficiency of the PV system, and characteristics of the PV system. The solar PV power profiles for each day of January, which is a summer season, were presented to show the variability of the solar PV power in numerical examples. The simulation results show relatively accurate forecasting with 17.57 kW and 2.80% as the best root mean square error and mean relative error, respectively. Thus, the proposed solar PV power prediction model can help power system engineers in generation planning for a grid-connected or stand-alone solar PV system.