Short-term wind power prediction of nuclear extreme learning machine based on improved northern goshawk optimization algorithm

Accurate wind power forecasting is crucial for optimizing wind resource utilization and ensuring grid safety. To enhance prediction accuracy, a new model is proposed that addresses the limitations of the Northern Eagle Optimization Algorithm (NGO), particularly in population initialization and local optimization. The improved model, called INGO, incorporates Cat chaotic mapping, adaptive inertia weight, and a longitudinal-transverse crossover strategy. Additionally, an optimized kernel extreme learning machine (KELM) is used. Experimental results show that INGO outperforms the standard NGO in terms of convergence speed, global search capability, predictive accuracy, and generalization, indicating strong potential for practical application.