A Hybrid Model For Short-Term Wind Speed Forecasting Based On Non-Positive Constraint Combination Theory

Short-term wind speed forecasting plays an irreplaceable role in efficient management of wind energy systems and accurate forecasting results could provide effective future plans for operators of utilities and wind energy systems. Aiming at improving the accuracy of short-term wind forecasting, this paper presents a new forecasting model based on the non-positive constraint combination theory. In this model, a modified optimization algorithm is used to optimize the weight coefficients of the constituent models based on the non-positive constraint combination theory. The combined model is tested using three sets of 10-min wind speed data from real-world wind farms. The testing results show that the forecasting accuracy of new model is significantly better than the constituent models.