PREDICTION OF SHORT TERM SOLAR ENERGY IRRADIANCE USING DEEP LEARNING, EXPLAINABLE AI, AND GENERATIVE AI: A COMPARATIVE STUDY OF ECONOMIC FEASIBILITY AND MODEL ACCURACY

Abstract

The authors of this paper explores the use of cutting-edge artificial intelligence (AI) technologies— specifically Deep Learning (DL), Explainable AI (XAI), and Generative AI (Gen AI)—for predicting solar energy irradiance. Given the critical role of accurate irradiance forecasting for optimizing solar energy production, this paper proposes a hybrid AI framework combining these technologies to improve prediction accuracy while ensuring economic feasibility. The proposed methodology integrates a DL-based model for time-series forecasting, with XAI techniques to enhance interpretability, and Gen AI to address data scarcity issues. We analyze the advantages and disadvantages of each technology in the context of solar energy prediction and provide recommendations on their practical deployment. The results show that the hybrid approach offers substantial improvements in forecasting accuracy and model transparency, making it a promising solution for real-world applications.