Akademik Veri Yönetim Sistemi
PROGRESS IN NUCLEAR ENERGY, cilt.195, ss.1-17, 2026 (SCI-Expanded, Scopus)
This study introduces an innovative methodological framework for nuclear power plant (NPP) site selection in Mersin Province, integrating the three best-performing Machine Learning (ML) algorithms from a set of seven, each offering distinct advantages, with Half Quadratic Programming (HQP) and Geographic Information Systems (GIS). The study achieved the most favorable outcomes with Categorical Boosting (CB), Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP). The integration of these algorithms, SVM (R2 = 0.960), MLP (R2 = 0.953), and CB (R2 = 0.912), significantly enhanced the accuracy and objectivity in determining criterion weights, thereby reducing reliance on subjective expert opinions. HQP effectively harmonized the weights derived from different ML methods, optimizing critical criterion interactions such as seismic risk, environmental factors, and infrastructure efficiency. GIS enabled precise spatial analyses, identifying coastal areas around Anamur and Aydıncık as most suitable (score 5), and inland regions near Gülnar and Silifke as highly suitable (score 4). Notably, the existing location of the Akkuyu NPP precisely matches the area identified as most suitable, validating the robustness and reliability of the proposed approach. Overall, this research significantly advances nuclear site selection methodologies, offering a comprehensive, objective, and practical decision-making tool that substantially contributes to strategic energy planning.