A data-driven multi-channel supply chain multi-factory collaborative production planning problem

Proper production planning is significantly for enhancing revenue and productivity of manufacturing companies. This paper focuses on multi-channel supply chain multi-factory collaborative production planning problem, considering uncertain customer demand. The decision-making process is divided into strategic and operational decisions aimed at maximizing company profits. To meet customer demand, the collaboration between multiple factories is crucial. However, customer demand is often uncertain, unobservable, and difficult accurately describe. To address this challenge, we propose a data-driven adaptive distributionally robust optimization model. Specifically, we utilize K-means clustering to tackle the unobservability of customer demand and construct a distributional ambiguity set described by divergence to handle the distributional ambiguity of customer demand. Subsequently, the formulated optimization model is transformed into a mixed integer program. In numerical experiments, we demonstrate the efficacy of our proposed adaptive distributionally robust optimization model in mitigating customer demand unobservability and distributional ambiguity. We verify the impact of cluster number on the model solution. Moreover, we compare our constructed adaptive distributionally robust model with the stochastic average approximation model, demonstrating that our model better reduces uncertainty compared to the stochastic average approximation model.