LAUSR

Abstract Wind energy is becoming increasingly competitive and promising for renewable energy profiles. Accurate and reliable wind speed prediction is crucial for the effective exploitation of wind energy. However, previous studies have generally ignored the importance of capturing both linear and non-linear wind speed characteristics and selecting forecasting sub-models objectively, resulting in poor forecasting performance. To bridge these gaps, a novel ensemble forecasting system is proposed by integrating the decomposition strategy, sub-model selection, and ensemble point and interval prediction based on the newly proposed multi-objective Archimedes optimization algorithm, which has been demonstrated to be effective at the theoretical and empirical levels for providing reliable wind speed forecasting results. Based on hourly-resolution wind speed data from three datasets of Shandong Peninsula, China, three experiments and discussions are conducted. Simulation results demonstrate that the proposed system is capable of obtaining a high degree of precision and reliability for both point and interval forecasting relative to other comparative models. Thus, it can provide credible references for power system dispatching and management.