| Abstract | The magnitude of power fluctuations at large offshore wind farms has a significant impact on the control and management strategies of their power output. These fluctuations can be regarded at different time scales. If focusing on the minute scale, one observes successive periods with smaller and larger power fluctuations. It seems that different regimes yield different behaviours of the wind power output. Therefore, the use of statistical regime-switching models is investigated here for better capturing these fluctuations. More precisely, Self-Exciting Threshold AutoRegressive (SETAR), Smooth Transition AutoRegressive (STAR) and Markov-Switching AutoRegressive (MSAR) models are considered. The particularities of these models are presented, as well as methods for the estimation of their parameters. Simulation results are given for the case of the Horns Rev and Nysted offshore wind farms in Denmark, for time-series of power production averaged at a 1, 5, and 10-minute rate. The exercise consists in one-step ahead forecasting of these time-series with the various regime-switching models. It is shown that the MSAR model, for which the
succession of regimes is represented by a hidden Markov chain, significantly outperforms the other models, for which the rules for the regime-switching are explicitly formulated. |