Optimal sizing of hybrid offshore wind-wave energy systems using neighborhood-reinforced ant colony optimization

• Novel method proposed for optimizing hybrid offshore wind-wave energy systems • Method outperforms classical sizing methods • Optimized hybrid system shows superior performance relative to single-source systems • Levelized Cost of Energy of hybrid offshore wind-wave energy system is competitive • Targeted policy support for wave energy can significantly reduce the LCOE For optimizing the size of hybrid renewable energy systems, researchers increasingly rely on artificial intelligence methods. In this work, a new Ant Colony Optimization variant is introduced: Neighborhood-Reinforced ACO (NR-ACO). Its key innovation lies in a neighborhood-based pheromone reinforcement mechanism that deposits pheromone not only on the identified best solution but also on nearby candidate solutions within a defined radius. This mechanism improves the exploration-exploitation balance in discrete sizing problems. This approach is tested on a hybrid wind-wave energy system – an underexplored configuration – designed for a remote Atlantic island: El Hierro (Spain). Across six test cases (three reliability levels and two locations), NR-ACO consistently outperforms Simple-ACO and Genetic Algorithm as benchmarks. The improvement can reach 11.7% in mean LCOE versus Simple-ACO, 8.9% versus GA, while also achieving about one order of magnitude lower standard deviation. The best configuration achieves an LCOE of $0.4037/kWh at 1% LPSP, corresponding to 17 floating 12-MW wind turbines, 315 wave energy converters rated at 400 kW and 2.03 GWh battery capacity. Sensitivity analysis shows LCOE is most responsive to wave energy capital cost. A 40% reduction in WEC CAPEX leads to an 18% LCOE decline, while a similar OPEX reduction results in only 3% LCOE decline. Findings indicate that offshore wind-wave hybrids are cost competitive for island systems and that NR-ACO provides a robust optimization approach for complex hybrid energy sizing problems.