Multidimensional LSTM Networks to Predict Wind Speed

Abstract

Wind speed forecasting has great significance to the improvement of wind turbine intelligent control technology and the stable operation of power system. In this paper, the Long Short-term Memory (LSTM) mode combined with deep learning ability which with the fuzzy-rough set theory has been proposed for short-term wind speed prediction. Using the fuzzy-rough set to reduce input and space features firstly. Then find the main factors that influencing the wind speed as the input for the LSTM neural network prediction model. Finally achieve the forecast of data trends through the strong generalization ability on massive of data learning. The experimental results show that the FRS-LSTM model has higher prediction accuracy than traditional neural network.