Short-term Wind Power Forecasting Based on Spatial Correlation and Artificial Neural Network

Abstract

Short-term wind power forecasting is essential for the efficient operation of power systems with high wind power penetration. A suitable power generation schedule can be made based on accurate wind power forecasting results. Artificial neural networks (ANNs) are capable of tracking the non-linear and complex wind speed patterns. Due to the availability of large amounts of historical data and strong computational power, ANNs have been a popular selection for non-linear forecasting in the last decade. Therefore, ANNs are good candidates for short term wind power forecasting. In this paper, spatial correlation between the target stations and the neighboring stations is investigated, and the meteorological variables from the most related three neighboring stations are used as input features to an ANNs with two hidden layers. The simulation results show that using the meteorological variables from the highly related neighboring stations can improve the short-term wind speed and wind power forecasting accuracy of the target stations.