This study addresses a multi-site stochastic order planning problem with real-world features existing in make-to-order textile manufacturing supply chains widely. A mathematical model of the problem is presented with the consideration of uncertain orders and heterogeneous carbon emission levels of different plants. The problem aims at minimizing the total tardiness and the excessive carbon emissions. By combining a single-objective-guided multi-objective evolution algorithm and a scenario generation technique, a multi-objective evolutionary stochastic optimization approach is proposed to solve this problem. For the first time, the single-objective-guided multi-objective evolutionary algorithm is adapted to handle discrete optimization problems. The iterative scenario selection-based scenario generation approach is used in evaluating the candidate solutions to the problem with uncertain orders. Extensive experimental results demonstrate the effectiveness of the proposed approach and the necessity of considering uncertain orders in real-world production planning.

本研究解决了一个多站点随机订单计划问题，该问题具有广泛存在于按订单生产的纺织品制造供应链中的真实世界特征。考虑到不同植物的不确定阶数和异质碳排放水平，提出了该问题的数学模型。该问题旨在最大限度地减少总延误和过多的碳排放。通过结合单目标引导的多目标进化算法和场景生成技术，提出了一种多目标进化随机优化方法来解决这个问题。单目标引导的多目标进化算法首次适用于处理离散优化问题。基于迭代场景选择的场景生成方法用于评估具有不确定顺序的问题的候选解决方案。大量的实验结果证明了所提出方法的有效性以及在实际生产计划中考虑不确定订单的必要性。