Machine Learning-enhanced optimization of Laser-based Direct Energy Deposition Additive Manufacturing Technology

Laser-based Direct Energy Deposition (DED-LB) is a key technology in both manufacturing and repair of components within metal Additive Manufacturing (AM). However, challenges such as heat accumulation and insufficient dynamic process control restrict its broader adoption. This work proposes a novel strategy to stabilize the deposition process by integrating a Machine Learning (ML) model as an online controller, coupled with a Finite Element (FE) solver for real-time process monitoring of the deposition process within a computational domain. The FE framework performs thermal analysis of the DED-LB process and generates synthetic data for training the ML model. The ML-enhanced control system then analyzes the melt pool morphology and thermal profiles from simulations in real-time, and outputs corrective laser power adjustments to maintain a constant penetration depth. This closed-loop system enables autonomous monitoring and rapid dynamic response, ensuring consistent thermal management throughout the deposition process. The ML model architecture was optimized through hyperparameter tuning and trained using synthetic data generated from high-fidelity simulations. Three scanning sequence scenarios are presented to evaluate the accuracy of the online monitoring system. Results demonstrate that integrating this control framework maintains a stable melt pool penetration depth, thereby enhancing geometric precision and reliability in DED-LB processes through tailored time-series power profiles. The generalization capability of the ML-based controller was demonstrated by its effective performance on scenarios beyond the training data. This result highlights its adaptability and improved process control compared to traditional parameter-tuned controllers. By improving dimensional accuracy and consistency of AM components, this study supports broader industrial adoption of the DED-LB technology. Additionally, it establishes a preliminary framework for evaluating the feasibility of the proposed control strategy, with the objective of future implementation for physical control of DED-LB machines by adjusting the laser power inreal-time.