Effect of Input Features on the Performance of the ANN-based Wind Power Forecasting

As the penetration of wind power increases in power system, accurate short-term wind power forecasts will become very crucial for effective grid operation and management. With the availability of a large amount of meteorological data and computational power, artificial neural networks (ANNs) have become a very popular method for forecasting. The purpose of this paper is to provide an inexpensive but reliable ANN model to forecast wind power produced from wind turbines for small wind farms. It is shown that the ANN model with all the input features and large training sample size has the best forecasting results compared to the one with small sample size and input features. The tradeoff between forecasting performance and computational cost is also analyzed in the paper.