Surrogate-Assisted Multi-Objective Snow Goose Algorithm With Gaussian Process for Cold Chain Logistics

This study presents a novel optimization framework, the Gaussian Process-Based Surrogate-Assisted Multi-Objective Snow Goose Algorithm (GP-SAMOSGA), aimed at optimizing cold chain logistics (CCL) by addressing dual objectives: minimizing transportation costs and maximizing user satisfaction. User satisfaction is quantified through two key dimensions—time window compliance, which reflects delivery punctuality, and product freshness preservation, which evaluates the maintenance of perishable goods’ quality. The optimization is conducted within essential operational constraints, such as temperature regulation, vehicle loading capacities, and specified delivery time windows. To address the computational complexity and high evaluation costs of the objective functions, the GP-SAMOSGA integrates a Gaussian Process (GP) surrogate model with the Multi-Objective Snow Goose Algorithm (MOSGA). The GP surrogate model approximates the expensive objective functions, reducing the number of true function evaluations. Meanwhile, the MOSGA utilizes the migration and foraging behaviors of snow geese to balance exploration and exploitation in the search space. This integration allows the GP-SAMOSGA framework to efficiently identify Pareto-optimal solutions and form a well-distributed Pareto front that captures the trade-offs between competing objectives. The performance of the GP-SAMOSGA framework is validated through experiments on benchmark problems and a real-world CCL case study. The results demonstrate the algorithm’s high solution quality, computational efficiency, and scalability. Moreover, the GP-SAMOSGA outperforms traditional multi-objective optimization algorithms, making it a promising tool for complex logistics optimization tasks. This research contributes to the field by combining surrogate modeling with bio-inspired optimization, offering a robust and efficient approach for multi-objective CCL optimization.