Abstract
This study addresses key limitations of traditional algorithms in green supply chain management, including vulnerability to local optima, unsystematic population initialization, and inadequate parameter tuning. To overcome these challenges, a collaborative optimization framework is developed by integrating an Improved Sparrow Search Algorithm (ISSA) with the Non-dominated Sorting Genetic Algorithm III (NSGA-III), forming the ISSA-NSGA-III model. The model begins with Tent chaotic mapping to generate a well-distributed initial population, ensuring broad coverage of the solution space. An adaptive periodic convergence factor is then introduced to dynamically balance global exploration and local exploitation during the optimization process. To further strengthen search performance, Lévy flight and elite opposition-based learning are incorporated, enhancing the model's ability to escape local optima. These improvements are embedded within the NSGA-III framework to achieve robust multi-objective optimization. Performance is evaluated using a public supply chain management dataset. Compared with the standard Sparrow Search Algorithm, which serves as its baseline, the ISSA-NSGA-III model reduces total supply chain cost by approximately 14.0%, lowers carbon emissions by 14.2%, and increases resource utilization by 15.4%. The spacing metric also improves by 40.2%, indicating a more uniform and diverse distribution of Pareto-optimal solutions. Overall, the proposed algorithm demonstrates strong potential for coordinating economic performance and environmental sustainability in green supply chains, offering reliable technical support for complex decision-making.