Surface roughness prediction of laser-based powder bed fusion manufactured parts using machine learning

Laser-based powder bed fusion in metal additive manufacturing faces industrial integration challenges due to low build-up rates. To overcome this, larger spot diameters, higher laser power, and faster scanning speeds can be used. However, these modifications affect the characteristic properties such as relative density, microhardness, and surface roughness of the final component. This work specifically looks at the influence of these process parameters on surface roughness and proposes a machine learning framework to predict surface roughness based on selected process parameters. Furthermore, it is demonstrated that the proposed framework can be used to optimize process parameter selection in the design phase.