Probability-scenario guided adaptive large neighborhood search for stochastic order allocation

In multimodal freight logistics the assignment of uncertain customer orders under carbon emission constraints presents a complex and important challenge. This study develops a high dimensional stochastic optimization model that aims to maximize the expected profit of order transportation by considering transportation costs and penalties caused by exceeding carbon limits. To solve this problem efficiently an intelligent optimization approach is proposed which integrates a probability guided adaptive large neighborhood search with a scenario generation technique. This method improves computational efficiency by identifying key scenarios and prioritizing influential order combinations during the search process. Experimental results indicate that the proposed approach yields improvements over the tested conventional methods in both solution quality and computational speed within the scope of our simulated scenarios. It demonstrates robust performance and stability in handling high-dimensional uncertainty, offering practical insights for sustainable logistics planning. The experimental findings indicate that the proposed method improves objective performance by over 10% on average while reducing computational time by more than 80% specifically when compared to the baseline random sampling-based intelligent neighborhood optimization algorithm used in this study. These results highlight the effectiveness of the approach in addressing high-dimensional stochastic logistics optimization under environmental constraints.