Wind Speed Prediction Using CNN-BiLSTM Network

Accurate wind speed prediction is quite imperative for the optimum performance and integration of wind energy into power systems. This paper presents a new approach in wind speed prediction by applying hybrid models from deep learning: Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory networks (BiLSTM), and Long Short-Term Memory networks (LSTM). The proposed method enhances prediction accuracy through the exploitation of spatial and temporal dependencies within wind speed data. First, an autoencoder model was used for imputation of missing values in the real-world dataset. Then, CNN was used to extract features, BiLSTM and LSTM networks were also inculcated capture sequential dependencies. Accordingly, the hybrid CNN-BiLSTM model outperformed other models about wind speed forecasting.