Stochastic Optimization Methodology for Production Planning with Uncertain Demand and Lead Time Based on the Digital Twin

Due to the manufacturing demand for mass customization and the widespread application of flexible automated production equipment, manufacturers tend to shorten their production cycle. This results in a loss of regularity in the production system, and it becomes difficult for manufacturers to control the product and process quality. To circumvent the issues related with these uncertainties, recent studies showed that using stochastic programming approach in Material Requirement Planning software can yield significant cost saving for the industry. However, there are very few studies that discuss uncertain production demand and lead time simultaneously. To fill these gaps, this paper considers the lot-sizing problems for production planning with uncertainties. We propose a stochastic optimization methodology for production planning with uncertain demand and lead time based on the digital twin. The proposed method is based on the simulation and Bayesian network forecast methods. In addition, we explain how the approach can be integrated with the digital twin systems. More precisely, we integrate the lot-sizing model with the simulation model, domain model, and Bayesian network model to forecast the lead time and improve the lot-sizing model with uncertain demand and lead time.