Symmetry-Guided Surrogate-Assisted NSGA-II for Multi-Objective Optimization of Renewable Energy Systems

Abstract

In this study, we propose a novel hybrid metaheuristic framework that leverages symmetry awareness to enhance the performance of multi-objective optimization in renewable energy systems. Motivated by the repetitive and structured patterns commonly found in solar, wind, and demand profiles, we introduce a symmetry-guided variant of the NSGA-II algorithm, enriched with a customized crossover operator that detects and exploits symmetrical patterns in candidate solutions. To further accelerate convergence and reduce computational cost, we integrate a surrogate modeling strategy using machine learning to approximate fitness evaluations in later generations. Our experimental evaluation, based on a synthetic dataset simulating one week (168 h) of operation in a hybrid solar–wind power system, incorporating realistic diurnal patterns in generation and demand, demonstrates the proposed method’s superiority over baseline NSGA-II in terms of solution diversity, convergence, and runtime efficiency. The results highlight the importance of integrating domain-specific structure—such as temporal symmetry—into the design of metaheuristics for sustainable energy applications. This approach opens new avenues for scalable, intelligent optimization in complex energy environments.