Wind Speed Prediction at Multiple Heights for Offshore Wind Turbines Using Deep Learning

Abstract

This work addresses short-term wind speed prediction in offshore wind farms, a key challenge for renewable energy transition. Short-term forecasting ensures stable grid operation. Applying Deep learning (DL) techniques faces computation time and forecast performance challenges. The paper proposes a practical approach using diverse DL models to forecast wind speeds at different heights. Geographically independent datasets from two continents are used for train the models and then test those. Four DL models namely Long Short Term Memory (LSTM), Bi-Directional LSTM (Bi-LSTM), Gated Recurrent Unit (GRU) and Convolutional Neural Network-LSTM (CNN-LSTM) are evaluated. Results are consistent and promising. In summary, the paper offers a practical solution for forecasting short-term wind speeds at multiple heights and time steps. It explores DL model variants, utilizes diverse datasets, and provides insights for effective offshore wind farm operations. As per our knowledge, such type of work on short term wind speed predictions at 10 heights up to 250 meters are not done yet.