Automating Process Optimisation from A Metrology Digital Twin

Manufacturing processes have variation of input conditions and environmental and operational disturbances such as material properties or equipment condition changing the behaviour of machining processes. This means that existing models cannot take the next step of in-process control without being dynamic, adapting to the specific conditions of each part. The aim of this project is to investigate new dynamic process models that act as a "digital twin" for forming and machining processes. Every part that passes through the manufacturing process will then carry its own unique set of model parameters derived from metrology data. The work will focus on part machinability, cutting forces, stability and residual stress distortions. Data will be monitored from forming and machining processes such as temperatures, forces, part dimensions and machining vibrations. This data will inform models that will be running in real time, driving predictions which in turn drive forming and machining parameters. The study will provide a strong case for further research into dynamic models that apply active automatic learning to optimise multi-stage manufacturing processes.