Key points are not available for this paper at this time.
Accurate and reliable short-term wind speed forecasting is essential for the effective utilization and scientific dispatch of wind energy. In recent years, with the rapid development of Artificial Intelligence (AI) technology, the capabilities of prediction models have been significantly improved. However, there are still many challenges in dealing with non-stationary signals, coordinating multi-objective conflicts, and enhancing model robustness. This paper reviews the three core links in wind speed forecasting: multi-scale signal decomposition, hybrid modeling framework, and multi-objective optimization algorithm. First, this paper systematically summarizes common signal decomposition methods, including Empirical Mode Decomposition (EMD), Variational Mode Decomposition (VMD), and wavelet transform-based techniques, and analyzes the characteristics of these methods in terms of frequency resolution, adaptability, and model ensemble capabilities. Secondly, from the perspective of structural form and functional complementarity, this paper explores how to combine shallow models, deep neural networks, and ensemble strategies to build an efficient hybrid prediction model. Finally, a variety of Multi-Objective Optimization (MOO) algorithms are evaluated, such as the Non-Dominated Sorting Genetic Algorithm II (NSGA-II), the Multi-Objective Particle Swarm Optimization Algorithm (MOPSO), and the Strength Pareto Evolutionary Algorithm II (SPEA-II). Their potential in model architecture adjustment, fusion strategy design, and balancing prediction accuracy, stability, and computational efficiency is analyzed. This paper comprehensively evaluates related studies according to the "decomposition-prediction-optimization" process. At the same time, it emphasizes the importance of developing a wind speed prediction system that is robust, interpretable, and easy to deploy. This study provides a unified perspective and reference framework for building intelligent and goal-oriented wind speed prediction solutions. • A comprehensive review on hybrid AI models for wind speed forecasting is presented. • Multi-scale decomposition techniques are compared in terms of adaptivity and performance. • Model fusion strategies are analyzed across shallow, deep, and ensemble learning. • Multi-objective optimization algorithms are evaluated for architecture and parameter tuning.
Liu et al. (Wed,) studied this question.