A Machine Learning Method To Quantitatively Predict Alpha Phase Morphology In Additively Manufactured Ti-6al-4v

This study presents a GAN architecture trained to synthesise microstructure images of LPBF fabricated Ti-6Al-4V based on different LPBF processing parameters. The ML model is trained on a multi-category image dataset of micrographs from samples fabricated with different laser powers and scan speeds. The model can accurately predict detailed microstructural features for LPBF processing ranges beyond the training dataset with a relative error of less than 20%. This approach can be widely applicable to solving problems of process optimisation and microstructure control in the broad field of additive manufacturing.