The Taïba Ndiaye wind farm, connected to the SENELEC grid, plays a key role in offsetting shortfalls in electricity consumption, with an installed capacity of 158.7 MW. Moreover, as an intermittent power station, its production is highly dependent on the environmental conditions in the region. Bad weather can disrupt the electricity network, requiring forecasting methods to anticipate its production. This will make it easier to decide how much fossil energy to bring on stream to meet demand. The aim of this paper is to provide forecasts of wind generation at Taïba Ndiaye, subdividing the data into 80% for model training and 20% to assess its robustness to generalization to other situations. The aim is to quantify the energy produced and facilitate an optimal transition between intermittent and fossil energy sources. Two artificial intelligence models classified as machine learning (decision tree and random forest) are proposed in the study, with respective coefficients of determination of 0.92 and 0.938. The results, compared with the literature, demonstrate the reliability of the approach using only production data. These results promise significant benefits in terms of resource management.

Taïba Ndiaye Wind power SENELEC grid forecast machine learning artificial intelligence models