LSTM Network for Forecasting Solar Energy Injected into the LV Electricity Grid: Case Study, Bokhol Solar Power Plant, Senegal

The integration of renewable energy sources into power grids represents a major challenge for optimizing the energy mix. Intelligent forecasting methods have become essential tools for improving energy services and supporting decision-making in production forecasting and efficient resource planning. This paper proposes an energy forecasting model based on the Long Short-Term Memory (LSTM) approach. The model is applied to data from the Ten Merina photovoltaic power plant in Senegal, which has an injection capacity of 20 megawatts sufficient to supply approximately 160,000 people, thus contributing significantly to the country's energy transition. The plant spans 46 hectares and comprises 92,000 polycrystalline glass module solar panels. The LSTM model demonstrates satisfactory performance, with a correlation coefficient of $\mathbf{9 4 \%}$, indicating a strong agreement between predicted and actual values