Long-term Power Generation Prediction in Photovoltaics Using Machine Learning-based Models

The research in the field of renewable energy has taken centre stage in the study of reliable and effective photovoltaic (PV) systems. These systems are essential to a future powered by renewable energy, where solar radiation is directly converted into electrical power. However, the photovoltaic arrays have limited conversion efficiency. Hence, highly accurate forecasting strategies are required to mitigate the impact of this challenge. This research focuses on proposing serial algorithms that combine machine learning and global optimization algorithms to solve stochastic optimization problems. Gated Recurrent Unit (GRU) architecture, Support Vector Machine (SVM) for Regression (SVR) models and Differential Evolution algorithm (DE) are used in developing the forecast of grid power generation across environmental variations. Initially, four serial GRU-SVR models will be trained to address the prediction for the seasonal evolution. Afterwards, a hybrid approach GRU-SVR-DE strategy will be defined to integrate four seasonal models, providing a robust forecasting strategy for PV power generation. In the end, the performances predictions will be analyzed to demonstrate the accuracy of the long-term forecasts.