Real-time multivariable control of directed energy deposition via adaptive model predictive control

Abstract

Additive manufacturing processes such as directed energy deposition (DED) enable precise material deposition and customization, but ensuring consistent material properties remains a challenge due to the complex interplay of process parameters. This research presents a novel adaptive model predictive control (AMPC) algorithm for real-time multivariable control in DED, integrating an adaptive one-dimensional thermal model for accurate prediction of both temperature distribution and spatial cooling rate. The model was experimentally validated in single-track deposition tests across four different materials, achieving temperature predictions within ±1% of infrared camera measurements and spatial cooling rate errors below 2.73%. The validated model was embedded within the control framework and evaluated in five-layer wall experiments under open-loop, single-input single-output (SISO), and multi-input multi-output (MIMO) closed-loop control configurations. Results demonstrate that the AMPC algorithm effectively stabilized melt pool dynamics through simultaneous control of laser power and traveling speed, leading to consistent layer heights and improved material uniformity. This work introduces a scalable, adaptive, physics-based framework for real-time thermal prediction and multivariable control in advanced manufacturing processes that use concentrated energy sources, improving melt pool stability, material consistency, and overall part quality.