A decision-making and planning optimization framework for implementing Hybrid Renewable Energy Systems in Mediterranean islesA decision-making and planning optimization framework for implementing Hybrid Renewable Energy Systems in Mediterranean islesΠλαίσιο λήψης αποφάσεων και βελτιστοποίησης σχεδιασμού για την διαστασιολόγηση υβριδικών συστημάτων ανανεώσιμων πηγών ενέργειας σε νησιά της Μεσογείου Μεταπτυχιακή Διατριβή Master Thesis 2025-07-242025enThis thesis presents an integrated, AI-enhanced decision-making and planning optimization framework for the deployment of Hybrid Renewable Energy Systems (HRES) in Mediterranean insular contexts. It addresses the multidimensional challenge of transitioning non-interconnected or partially interconnected island grids toward energy autonomy, economic viability, and deep decarbonization. The work introduces a robust methodology that combines energy demand profiling, stakeholder-informed decision criteria, and AI-driven weight extraction using Random Forest regression, SHAP values, and the CRITIC method. The framework simulates and evaluates 14 distinct system configurations, integrating photovoltaic (PV), wind, tidal, and wave resources with energy storage solutions, via a custom Genetic Algorithm (GA) model. The GA incorporates a novel adaptive mutation and restart mechanism to escape local optima, yielding convergence-optimized solutions tailored to real insular constraints. Scenario ranking is performed using a multi-criteria decision-making approach based on the TOPSIS method, informed by AI-extracted criteria weights. Comparative analysis of weighting strategies reveals consistent identification of the most suitable configuration, validating the robustness of the proposed approach. The model is designed to be modular, adaptable to diverse island conditions, and fully transparent in its evaluation of trade-offs across economic, environmental, social, political, and technical dimensions. Validation is achieved through a case study on Crete, incorporating stakeholder feedback, dynamic load patterns, and renewable potential data. The resulting framework demonstrates its capacity to inform sustainable energy planning with scientific rigor and operational relevance. Lastly, this research contributes to filling a documented gap in HRES optimization literature, namely the absence of integrated, data-driven, stakeholder-informed MCDM models tailored to the insular energy context. The proposed framework advances methodological practices for sustainable energy transition planning and provides a replicable tool for decision-makers engaged in island greenification and resilience strategies. http://creativecommons.org/licenses/by-nc-nd/4.0/Πολυτεχνείο Κρήτης::Σχολή Μηχανικών Παραγωγής και ΔιοίκησηςSifakis_Nikolaos_MSc_2025.pdfChania [Greece]Library of TUC2025-07-24application/pdf17.3 MBembargo Sifakis Nikolaos Σηφακης Νικολαος Zopounidis Konstantinos Ζοπουνιδης Κωνσταντινος Doumpos Michail Δουμπος Μιχαηλ Kontogiannis Thomas Κοντογιαννης Θωμας Πολυτεχνείο Κρήτης Technical University of Crete Hybrid Renewable Energy Systems Sustainability in Islands Multi-Objective Optimisation AI-enhanced Multi-Criteria Decision Making