Offshore wind energy is deeply coupled with the characteristics of offshore wind conditions. The intermittent characteristics of wind speed and its spatiotemporal fluctuation pose a serious challenge to the safety and stability of offshore wind power grid-connected systems, which makes the accuracy of offshore wind speed prediction a key factor affecting the development and utilization of offshore wind energy. To address the limitations of conventional prediction models in handling noise interference and spatiotemporal drift phenomena inherent to traditional methodologies, this study innovatively proposes a quadratic decomposition algorithm (SSA-EMF-SGMD) and establishes a hybrid prediction model (SSA-EMF-SGMD-PatchMixer) through the integration of singular spectrum analysis (SSA), entropy mean filtering (EMF), symplectic geometric mode decomposition (SGMD), and patchmixing network (PatchMixer). This study selects three regionally representative offshore wind farm observation Data sets and carries out the comparison experiment of multistep (1/4/8/12) wind speed deterministic prediction and probabilistic prediction. The results show that in deterministic prediction, the proposed model shows significant performance advantages compared with the baseline model, and the root mean square error is reduced by at least 26.5%. In probabilistic prediction, the model exhibits a stable coverage rate close to 90%, which verifies the reliability of the prediction results. In addition, the multistrategy training scheme constructed for actual engineering applications verifies the engineering applicability of the model in complex scenarios, which provides a highly reliable solution for offshore wind power grid-connected regulation.
Li et al. (Wed,) studied this question.