State transition ANNs for hourly wind speed forecasting

Accurate wind speed prediction is vital for improving the efficiency and reliability of wind power system operation. In this paper, a novel structure of forecasting method, the State Transition ANN model (T-ANN), is proposed for hourly wind speed forecasting. In the proposed method, the `state' is determined to describe the characteristic of the wind speed time series. Mapping relationship between two states is modeled using Artificial Neural Networks (ANNs) in the proposed method. Real world wind speed dataset is adopted to evaluate the proposed TANN wind speed forecasting model. Forecasting results show that the proposed T-ANN model outperforms the traditional regression ANN model (R-ANN) and the Support Vector Regression (SVR) model in 1 to 3 step wind speed forecasting.