Digital twin-driven multi-variable process control of thermal manufacturing processes

Thermal manufacturing processes are complicated to the extent that they require sophisticated performance optimization. However, their implemented form often lacks advanced control strategies. This research proposes digital twins (DTs) that generalize the concept of process control towards multivariable performance optimization. To achieve this, data-driven models are used to describe the mathematical link between the process parameters and the performance indicators (KPIs), while they are all defined based on the spatiotemporal profile of the temperature field. The control strategy is then designed based on time and quality criteria, while the control enforcement is regarded in real-time manner, so that the DT concept is validated. The performance of the process models is examined through aggregation of modelling details from physics-based models from an LPBF process, including delays and dynamics. The current proposed framework seems to close the gap between the theoretical and data driven models with respect to process modelling and control efficiency.