A Digital Twin-Driven Methodology for Material Resource Planning Under Uncertainties
Abstract
With the Industry 4.0 revolution currently underway, manufacturing companies are massively adopting new technologies to achieve the virtualization of their shop floor and the collaboration of their information systems. This process often leads to the construction of a real-time, collaborative, and intelligent virtual factory of their physical factory (so-called digital twin). The application of digital twins and frontier technologies in production planning still faces many challenges. But the research is still limited about how these frontier technologies can be applied to enhance production planning. This paper introduces how to enhance material resource planning (MRP) with digital twins and other frontier technologies, and presents a framework for the integration of MRP software with digital twin technologies. Indeed, the data collected from the shop floor can improve the accuracy of the optimization models used in the MRP software. First, several MRP parameters are unknown when planning, and some of these parameters may be accurately forecasted from the data with machine learning. Nevertheless, the forecast will never be perfect, and the variability of some parameters may have a critical impact on the resulting plan. Therefore, the optimization approach must properly account for these uncertainties, and some methods must allow building probability distribution from the data. Second, as the optimization models in MRP are based on aggregated data, the resulting plans are usually not implementable in practice. The capacity constraints may be acquired by communication with an accurate simulation of the execution of the plan on the shop floor.
Origin | Files produced by the author(s) |
---|