Very Short Term Wind Speed Forecasting Using Convolutional Long Short Term Memory Recurrent Neural Network

Abstract

To control, manage and optimize the generation of electricity form wind resource, an accurate forecasting of wind speed is imperative. This paper illustrates the implementation of a Convolutional Long Short Term Memory (CLSTM) neural network to forecast the wind speed on a half hourly basis for up to six steps ahead. To harvest maximum privilege form the uncertain wind energy, both point wise and probabilistic forecasting approach are combined together in this research. Two and a half years of historic dataset of wind speed and related variables (temperature and relative humidity) are used to train the model by dividing them into three cases. The performance of the proposed model is evaluated in terms of Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE) and Correlation between actual and predicted values. Besides, the performances of the proposed model is also compared with deep learning based Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM) Neural Network models. Results from the experiments shows that the proposed model has forecasted the wind speed with a better accuracy.