Machine Learning Based Models for Photovoltaic Energy Forecasting.

Abstract

Photovoltaic (PV) technology is one of the most promising forms of renewable energy, with its integration into the power grid increasing daily due to its economic and environmental ad-vantages. However, power generation from PV technologies is highly dependent on weather conditions, which are neither constant nor controllable. Therefore, accurate forecasting of PV power is essential for maintaining stability and reliable operation within the electrical power system. The goal of this project is to analyze and compare various machine learning-based forecast-ing methods based on their characteristics and performance. We utilized large datasets of measured PV power and meteorological parameters, such as solar radiation, temperature, and wind speed, which influenc eenerg ygeneration .Specifically, we proposed a machine learning-based model to forecast regional PV power for application in the Algerian energy market. Experimental results have shown that the Artificia lNeura lNetwork s(ANN )mode lexcels in capturing intricate linear and non-linear interactions of the input features, making it the most effective in forecasting solar generation.