System Identification and Feedback Control for Directed-Energy, Metal-Based Additive Manufacturing

Additive manufacturing of metal parts is a complex process where many variables determine part quality. In addition to manipulated process variables, such as travel speed, feedstock flow pattern, and energy distribution, other exogenous inputs also determine part quality. For example, changing build geometry and a growing global temperature. In addition, there are random external disturbances such as spatter on a cover lens. Both manipulated process variables and exogenous inputs affect dimensional tolerance, microstructure, and other properties that determine the final part quality. Our long term aim is to improve part quality through real-time regulation of measurable process variables using vision-based feedback control. As a starting point, we present a process model that relates scanning speed and laser power to build height and melt pool width. These results demonstrate the necessity for using multi-input multi-output feedback control techniques and provide information for refining the frame rate and spectral sensitivity of the imaging system.