Solar irradiation intensity forecasting for solar panel power output analyze

Abstract

Accurate forecasting of global horizontal irradiance (GHI) is critical for optimizing solar power plant (SPP) output, particularly in tropical locales where solar potential is high yet underutilized due to forecasting challenges. This research aims to enhance GHI prediction in one of the major cities of Indonesia, where existing models struggle with the area’s natural climate unpredictability. Our analysis harnesses a decade of data 2011-2020, including GHI, temperature, and the Sky Insolation Clearness Index, to calibrate and compare these methodologies. We evaluate and contrast the exponential smoothing method versus the more complicated artificial neural network (ANN). Our findings reveal that the ANN method, notably its fourth iteration model with 12 input and hidden layers, substantially outperforms exponential smoothing with a low error rate of 1.12%. The use of these methodologies forecasts an average energy output of 252,405 Watt for a solar panel with specification 15.3% efficiency and 1.31 m² surface area throughout the 2021 to 2025 timeframe. The work offers the ANN method as a strong prediction tool for SPP development and urges a further exploration into more advanced forecasting methodologies to better harness solar energy.