A Hybrid Approach for Short-term Wind Speed Prediction in Huan County of China

Abstract

Wind energy, which is intermittent due to the irregular and non-stationary characteristics of wind speed, can have a significant impact on power grid security. It is important to improve the accuracy of wind speed forecasting models for the wind generation. However, due to the nonlinear and intrinsic complexity of weather parameters, it is difficult to predict wind speed accurately by using different patterns in different locate. In this paper, a new hybrid wind speed forecasting model is constructed based on a back-propagation neural network(BPNN) and the idea of eliminating noise effects by using ensemble empirical mode decomposition(EEMD) method and eliminating seasonal effects from actual wind speed dataset using seasonal exponential adjustment(SEA). The hybrid EEMD-SEA-BPNN models are proposed to forecast the wind speed effectively in Huan County of Loess Plateau in China; numerical results demonstrate that the hybrid EEMD-SEA-BPNN model has better forecasting performance.