Amid escalating climate change and energy crises, wind energy, as a pivotal renewable resource, poses significant challenges to grid stability and energy management due to its inherent stochastic intermittency and nonlinear dynamics. Consequently, this research presents a hybrid prediction system, ICEEMDAN-NCRBMO-AELM, integrating data decomposition with intelligent computing to reveal spatiotemporal coupling patterns in climatic variables for reliable wind power forecasting. This system utilizes Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) to dissect sequences into several modes, addressing time–frequency features and alleviating mode mixing through a dynamic noise-weighting scheme. To further optimize Adaptive Extreme Learning Machine (AELM) performance, this research proposes a novel Normal Cloud Red-billed Blue Magpie Intelligent Optimization (NCRBMO) algorithm, motivated by cloud model theory and the swarm behavior of Red-billed Blue Magpie. NCRBMO employs a multiphase mapping inverse generation strategy for initializing individuals and designs five heuristic search strategies for global optimization. Regarding hyperparameter tuning, NCRBMO optimizes the weight matrix and bias vector in the output layer of a single-hidden-layer feedforward network, enhancing prediction accuracy and stability. The interseasonal wind power prediction results from Jiangsu region, China, indicate that this system surpasses competing representative techniques in addressing complex seasonal trends and meteorological abrupt changes.
Liu et al. (Sun,) studied this question.