A two-stage fuzzy nonlinear combination method for utmost-short-term wind speed prediction based on T-S fuzzy model

Wind speed prediction is a complex task in the field of wind energy resource forecasting. For prediction, highly nonlinear temporal features are required for wind speed time series, which are highly variable data sources. In this paper, a two-stage fuzzy nonlinear fusion model is proposed for the utmost short-term wind speed prediction problem of 5 and 15 min ahead. First, empirical mode decomposition decomposes the wind speed time series, and the resulting intrinsic mode functions (IMFs) are employed as features in the later modeling study. The first stage of modeling follows. Each IMF feature is fed into one of the three sub-models of the T-S fuzzy model based on triangle, fuzzy C-mean clustering, and Gaussian, yielding three prediction outputs. The second stage is then modeled, which takes advantage of the IT2-based nonlinear aggregation mechanism to overcome the inherent flaws of single methods and linear combinations. Finally, two real cases from wind farms in Colorado, USA, are analyzed to demonstrate the validity of the TFG-IT2 model. The prediction effect of various approaches was measured using three assessment indicators and a statistical test. The simulation results reveal that the TFG-IT2 model outperforms the other seven models in terms of prediction accuracy.