Deep Learning Model with Probability Density Function and Feature Engineering for Short Term Wind Speed Prediction

Abstract

Wind speed prediction in a given location is crucial for the evaluation of the wind power project. The accurate prediction of the wind speed improves the planning, reduces the cost, and improves the use of resources for wind power generation. There has been growing interest in the field of deep learning and neural networks for the prediction of wind speed as it can overcome the issue of accurately forecasting the nonlinear patterns of wind speed data using classical time series methods. In this study, a weather data set provided by the National Oceanic and Atmospheric Administration of the South Plains Region and offshore region of Texas is investigated for short-term wind speed prediction. At first, a probability density-based model and a simple linear model are applied for wind speed prediction. Then, the various feature has generated that impact the wind speed prediction. After that, various advanced deep learning models such as Convolutional and Recurrent Neural Networks (CNNs and RNNs) are built and applied to the data set obtained after feature engineering to predict the wind speed. The performance of nine different hybrid models used in this study is compared using different methods to evaluate the accuracy such as Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE). Based on the accuracy the best model is chosen and proposed for the wind speed prediction of the regions. The study shows the superiority of Encoder-Decoder LSTM over other hybrid deep learning models.