Real-Time Adaptive Layered Laser Sintering Process Optimization via Reinforcement Learning and Bayesian Neural Networks

**Abstract:** This research proposes a novel framework for real-time optimization of Direct Metal Laser Sintering (DMLS) processes, focusing specifically on the microstructural uniformity of Inconel 718 components fabricated via powder bed fusion. Leveraging Reinforcement Learning (RL) and Bayesian Neural Networks (BNN), the system dynamically adjusts laser power, scan speed, and scan strategy based on in-situ thermal feedback and predictive microstructure simulations. This approach aims to surpass existing reactive control methods and achieve unparalleled process control and microstructural homogeneity, resulting in superior mechanical properties for critical aerospace applications. It demonstrates immediate commercial viability via enhanced productivity and reduced scrap rates within existing DMLS production workflows. --- *This document is a part of the Freederia Research Archive. Explore our complete collection of advanced research at [en.freederia.com](https://en.freederia.com), or visit our main portal at [freederia.com](https://freederia.com) to learn more about our mission and other initiatives.*