Comparative Analysis of LSTM, GRU, and MLP Neural Networks for Short-Term Solar Power Forecasting

Renewable energies, including solar energy, have gained significant importance in the global energy landscape as a sustainable and eco-friendly alternative to fossil fuels. Solar energy, harnessed from solar radiation and meteorological factors, plays a crucial role in power grids worldwide. Improving energy output forecasts is essential for efficient planning and optimization of solar energy generation in the future.The primary objective of this research was to create and assess three distinct neural networks, namely Long-Short-Term-Memory (LSTM), Gated Recurrent Unit (GRU), and Multi-Layer Perceptron (MLP) for predicting the power generation of PV solar system. The training data was sourced from Adrar City, Algeria, covering the period from 2016 to 2020. The examination of these algorithms' performance revealed that LSTM neural network outperformed both GRU and MLP models. This can be attributed to LSTM's adeptness in managing time-series data, which significantly enhanced its ability to accurately forecast solar power production.