Passivity-based Iterative Learning Control Design for Selective Laser Melting

Selective laser melting (SLM) is an additive manufacturing process that creates 3D parts through layer by layer melting and fusion of a metal powder bed. Although a number of finite element models (FEM) have been developed that describe the coupled and complex physics associated with this process, they are typically not suitable for control algorithm design. In this manuscript, a control oriented reduced order model (ROM) to adequately capture these temperature dynamics is proposed and validated against high fidelity FEM simulations. Further, since the laser paths are often repetitive (or consist of repeating sub-trajectories), iterative learning control (ILC) algorithms can be used to obtain suitable laser power profiles to deliver desired temperature field profiles. However, the process is inherently multiple input single output (MISO), therefore, a suitable output is constructed in such a way so as to the make the system passive. A passivity-based ILC law is then designed to drive this synthesized output to a desired profile and a convergence criterion for this law derived. The proposed ILC update law is implemented on both models and the results are compared for a set of candidate laser paths. Finally, the ILC update law is implemented on the high-fidelity FEM to melt a ring geometry to demonstrate the capability of the ILC algorithm to generate optimal laser power profiles for creating complicated geometries on large powder beds.