Supply network (SN) robustness has become an important issue in SN management. In this paper, we refer to SN as robust, if it maintains its performance in the presence of uncertainty in SN parameters, namely uncertain changes in customer demand. A customer forecasts its demand in terms of requested quantity and time of delivery. This forecasted demand can be changed until a certain time. After that, the customer is committed to its demand. However, a manufacturer has to order materials in advance to produce its product without knowing the exact changes in customer demand. The materials can be ordered either from a standard supplier, or, from an emergency supplier, if there is not enough material in stock and/or there is not enough time for a delivery from the standard supplier. We define a new concept of fuzzy scenarios that comprise uncertain changes in customer demand. These changes are specified by linguistic terms and modelled using fuzzy numbers. The robustness of an SN is measured in a novel way as the variance of costs incurred in all fuzzy scenarios. This means that the robust SN maintains its cost in the presence of uncertain changes in customer demand. A novel fuzzy multi-objective optimisation model is developed, which determines quantities of materials to be ordered by a manufacturer from a standard supplier and times of ordering. The objectives considered simultaneously embed all fuzzy scenarios and include the minimisation of total SN cost, the maximisation of robustness and the minimisation of shortages. Various experiments are carried out to analyse the relationship between SN parameters and SN performance. Results obtained by applying the SN model demonstrate that robustness can be increased and shortages can be decreased, but, as expected, at a higher SN cost. In the case of the high ratio of the unit purchase cost from the emergency supplier to the unit surplus cost, a considerable increase of robustness and a decrease of shortages can be achieved. Finally, it is shown that the model can be applied to large-scale SNs.

供应网络 (SN) 稳健性已成为 SN 管理中的一个重要问题。在本文中，如果 SN 参数存在不确定性（即客户需求的不确定变化）时仍能保持其性能，我们将 SN 称为鲁棒性。客户根据要求的数量和交货时间预测其需求。这个预测的需求可以改变，直到某个时间。之后，客户致力于满足其需求。但是，制造商必须提前订购材料以生产其产品，而无需了解客户需求的确切变化。如果库存不足和/或没有足够的时间从标准供应商处交货，则可以从标准供应商处或从紧急供应商处订购材料。我们定义了一个模糊场景的新概念，其中包含客户需求的不确定变化。这些变化由语言术语指定并使用模糊数建模。SN 的鲁棒性以一种新颖的方式衡量，作为所有模糊场景中发生的成本的方差。这意味着在客户需求发生不确定变化的情况下，稳健的 SN 保持其成本。开发了一种新的模糊多目标优化模型，该模型确定制造商从标准供应商处订购的材料数量和订购次数。考虑的目标同时嵌入所有模糊​​场景，包括总 SN 成本的最小化、鲁棒性的最大化和短缺的最小化。进行了各种实验来分析SN参数与SN性能之间的关系。通过应用 SN 模型获得的结果表明，可以增加稳健性并减少短缺，但正如预期的那样，以更高的 SN 成本。在紧急供应商的单位采购成本与单位剩余成本的比率较高的情况下，可以实现稳健性的显着提高和短缺的减少。最后，表明该模型可以应用于大规模 SN。可以实现稳健性的显着提高和短缺的减少。最后，表明该模型可以应用于大规模 SN。可以实现稳健性的显着提高和短缺的减少。最后，表明该模型可以应用于大规模 SN。