A Comparative Study for Short Term Wind Speed Forecasting using Statistical and Machine Learning Approaches

The paper presents a comparative study of statistical and machine learning based methods for wind speed forecasting. Effectively predicting wind speed forecasts plays a vital role in resource planning and managing production control. Already existing forecasting methods are compared with the new ones over two different wind farm zones for short term forecasting horizon. Three most accurate and accessible methods Support Vector Regression (SVR), Auto-Regressive Integrated Moving Average (ARIMA), and Recurrent Neural Network are discussed along with the procedure for selecting correct model parameters for fine-tuning. A novel work on Kernel variations in support vector regressors is done to improve forecast results.