Quantile Regression-Based Solar Power Prediction

Abstract

Recent decades have seen an increase in the market for sustainable and green energy solutions. In all renewable resources, solar energy is one of the key sources because of its abundance. Integration of solar power into smart grids for meeting expanding energy demand relies on accurate predictions of solar power output. The efficiency of photovoltaic (PV) systems is highly influenced by climatic conditions, resulting in fluctuating power output that degrades the performance of smart grid systems. Hence, this work provides a probabilistic machine learning approach for solar power forecasting using reliable daily solar generation data. This work compares the effectiveness of predictive algorithms to forecast short-term solar power. The Normalized Mean Squared Error (NMSE) metric has been employed to assess the predicted accuracy of these models. The simulations and results highlight the effectiveness of probabilistic models, offering insights into enhancing grid stability and boosting the use of renewable energy sources.