Resilient agricultural biomass supply chain network design with uncertain disruptions: a data-driven globalised robust optimisation approach
针对农业生物质供应链易受不确定中断影响的问题，提出一种数据驱动的两阶段全局化鲁棒优化方法，通过随机森林构建不确定性集，平衡鲁棒性与经济性，并以湖南省秸秆供应链案例验证其有效性。
Agricultural biomass supply chain is highly vulnerable to uncertain disruptions, which threaten efficient utilisation and sustainable development. Designing a resilient biomass supply chain network is crucial to address these challenges. This paper investigates the agricultural biomass supply chain network design (BSCND) problem with uncertain disruptions. A novel data-driven two-stage globalised robust optimisation (TGRO) approach is proposed, which integrates pre-disruption and post-disruption decisions and constructs a worst-case mean-CVaR objective with ϕ-divergence distance to capture risk-averse preferences. Specifically, the data-driven inner and outer uncertainty sets are developed by random forest algorithm to characterise typical and extreme disruption probabilities, enabling a flexible balance between robustness and economic performance. The resulting semi-infinite TGRO model is reformulated as an equivalent mixed-integer linear programming model to ensure computational tractability. Finally, the effectiveness of the proposed TGRO approach is illustrated by a real-world case study on the straw supply chain network in Hunan province of China. The numerical results demonstrate that the proposed TGRO model outperforms both stochastic optimisation and classical robust optimisation approaches. It achieves a significant improvement in network robustness without falling into excessive conservatism, thereby ensuring greater post-disruption profitability and more stable decisions in facility location and flow allocation under different disruption scenarios.