Université du Québec à Rimouski

Published purpose

The project aims to increase metallographic laboratory autonomy and facilitate the development and optimization of processes, such as high-pressure vacuum casting (HPVDC) of aluminum, cold spraying of aluminum (CSAM) and steel made by selective laser melting (SLM), by applying generative artificial intelligence (AI) models. The project will focus first on increasing efficiency by quickly and cost-effectively generating machine learning datasets with adjustable features, making it possible to generate aluminum and steel microstructures with variations in process parameters and sample preparation conditions. Next, the application of models trained on generated images to increase the efficiency and autonomy of metallographic laboratories will be investigated, overcoming the scarcity of out-of-distribution image data by reconstruction using generative AI models. Finally, the use of AI models to reconstruct microstructure images and extract new features will be explored, with the aim of improving the prediction of mechanical properties. These developments could eventually lead to a product implemented by companies specializing in computer vision.