Reinforcement Learning-Based Feedforward Control for Solidification Cooling Rate Regulation in Directed Energy Deposition

Directed Energy Deposition (DED) is a widely used additive manufacturing for fabricating metal components with complex geometry and mechanical properties. A critical challenge in DED is the regulation of the melt pool's solidification cooling rate (SCR), which directly influences the microstructure and mechanical properties of the fabricated part. This paper proposes reinforcement learning (RL)-based feedforward control for the SCR regulation in DED. An RL control is developed to optimize process parameters, namely, laser power, traverse speed, and inter-layer dwell time, on a layer-by-layer basis. The proposed method leverages offline finite difference simulations to iteratively learn an optimal control policy that minimizes SCR deviations from a reference value while reducing overall production time. The effectiveness of the RL controller is evaluated for different references, demonstrating superior regulation performance and production time compared to control with constant process parameters.