Day Ahead Wind Energy Forecasting in Two Distinct Locations of Bangladesh Using Deep Learning Models

Bangladesh is looking for accurate wind energy projections in order to lessen its reliance on fossil fuels and diversify its energy sources. Proper forecasting increases wind farm efficiency, which improves stability of the grid and fosters economic expansion in the developing field of renewable energy in the nation. Using cutting-edge deep learning models, this study forecast day ahead wind power generation in Parki and Sitakundu, Bangladesh using Convolutional Neural Networks with bidirectional gated recurrent units (CNN-BiGRU), CNN-BiLSTM (Bidirectional Long Short-Term Memory), and LSTM-GRU. The models were trained on data from 2020-2022, including factors such as time, humidity, height, power, and barometric pressure, to forecast wind energy one day ahead. Among the models, CNN-BiLSTM was identified as the most effective wind energy for forecasting in both Parki and Sitakundu. It outperformed others with the lowest error rates, achieving Mean Absolute Error (MAE), Mean Square Error (MSE), and Root Mean Square Error (RMSE) values of 0.41, 0.42, and 0.65, respectively, in both locations. In contrast, the LSTM-GRU model had the least accurate results, with higher error values, including MAE of 0.58 and MSE and RMSE of 1.01 each.