Regime-switching modelling of the fluctuations of offshore wind generation

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. If focusing on the minute scale, it looks like different regimes yield different behaviours of the wind power output. The use of statistical regime-switching models is thus investigated. Regime-switching approaches relying on observable (i.e. based on recent wind power production) or non-observable (i.e. a hidden Markov chain) regime sequences are considered. The former approach is based on either self-exciting threshold autoregressive (SETAR) or smooth transition autoregressive (STAR) models, while Markov-switching autoregressive (MSAR) models comprise the kernel of the latter one. The particularities of these models are presented, as well as methods for the estimation of their parameters. The competing approaches are evaluated on a one-step ahead forecasting exercise with time-series of power production averaged at a 1, 5, and 10-min rate, at the Horns Rev and Nysted offshore wind farms in Denmark. For the former wind farm, the one-step ahead root mean square error (RMSE) is contained between 0.8% and 5% of installed capacity, while it goes from 0.6% to 3.9% of installed capacity for the case of Nysted. It is shown that the regime-switching approach based on MSAR models significantly outperforms those based on observable regime sequences. The reduction in one-step ahead RMSE ranges from 19% to 32% depending on the wind farm and time resolution considered. The presented results clearly demonstrate that the magnitude of fluctuations of offshore wind power cannot be considered as simply influenced by the generation level only.