Prediction of Wind Speed using Random Forest Learning Model considering multi-meteorological Variables

Wind power and other renewable energy sources are crucial for minimizing emission of greenhouse gas and preventing weather variation. The prediction of wind speed accurately is essential for maximizing production of wind energy and successfully integrating it into the electrical grid. In this paper, the real time data of wind speed for 24 hours of Bengaluru city (Karnataka, India) from 2011-2021 i.e., 11 years is considered to train Ensemble Learning based Random Forest (RF) learning model to forecast wind speed. The meteorological variable quantities such as dry bulb temperature, relative humidity and dew point are correspondingly taken as input parameters to train RF model. The short-term prediction of the wind speed is estimated for the day ahead and one hour. The statistical measures such as mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE) and normalized root mean square error (NRMSE) are evaluated to assess the performance of the results obtained from RF model. The outcomes determined from the proposed RF model is also equated with the conventional and advanced method to evaluate its efficacy and accuracy. The analysis demonstrate that the Random Forest model provides improved outcome than existing methods, providing accurate and reliable wind speed forecasts by handling non-linearities in the data.