Designing a cloud logistics mathematical model for photovoltaic systems using a metaheuristic algorithm

Abstract

This research develops and implements a cloud logistics model for solar photovoltaic (SPV) systems, optimizing the supply, distribution and management of electrical energy in the photovoltaic supply chain using deep learning and metaheuristic methods. It lays a foundation for future data-driven studies aimed at improving logistics performance in SPV systems. To validate the model, it was tested at randomly selected decision points, and its effectiveness was assessed through numerical examples with calculated objective function values. Experiments were conducted across small, medium and large dimensions to compare deterministic solutions with those from the NSGA-II metaheuristic algorithm. Results showed that as problem dimensions increased, complexity and solution times rose for both methods, with the NSGA-II algorithm significantly outperforming the deterministic approach in terms of speed. Additionally, nondominated points from the epsilon-constraint method were presented for demand increases of 10%, 20% and 30%, demonstrating convergence along established boundaries. This study proposes a novel cloud logistics model for SPV systems by integrating deep learning with metaheuristic optimization, particularly NSGA-II, to improve supply, distribution and energy management decisions in photovoltaic supply chains. Unlike prior studies that mainly address general sustainability, blockchain-based traceability or closed-loop logistics separately, the proposed model provides a unified data-driven decision-support framework and is validated across different problem scales.