A STUDY OF THE LASER POWDER BED FUSION MANUFACTURED SURFACE ROUGHNESS PREDICTION AND OPTIMIZATION BASED ON ARTIFICIAL NEURAL NETWORK

Laser powder bed fusion (L-PBF) is one of the most popular additive manufacturing methods in metal additive manufacturing. The surface roughness of the L-PBF manufactured parts is significantly influenced by major process parameters adopted, such as laser power, laser scanning speed, laser hatch space, and layer thickness of powder. This paper presented a machine-learning approach to determine the control parameters for reducing the surface roughness of the L-PBF products. In this study, a three-level full factorial experiment was designed with four major process parameters to examine the corresponding influence of surface roughness. The roughness of the fabricated samples was acquired by a laser profilometer. An artificial neural network model was trained to establish the relationship between the controllable parameters and finished surface roughness. The trained model was validated through additional experiments. The study provided an effective model for predicting the surface quality of L-PBF fabricated parts and a guideline for the process parameter selection.