Toward a Spark Plasma Sintering Digital Twin using a Deep Learning-Based Approach

Authors:

Damien Sicard (1,2), Mostapha Ariane (2), Frédéric Bernard (1), Foad Naimi (2)

1- ICB, UMR 6303 CNRS/Université de Bourgogne, Dijon, France

2- SINTERMAT SAS, Venarey Les Laumes, France

Abstract:

Over the last two decades, Spark Plasma Sintering (SPS) has become a major technique for manufacturing advanced materials. Nevertheless, the control of SPS process is complex and requires the use of complex multiphysics and multiscale numerical simulations. Nowadays, the emerging data-driven approaches such as Deep Learning (DL) have proven their effectiveness in many fields. Thus, we develop a DL architecture based on Convolutional Neural Network (CNN) and Generative Adversial Neural Network (GAN). The network is trained on high-throughput macroscale FEM simulation maps and associated process parameter tabular data. The power of this approach lies in the ability of the network training process to be incrementally augmented by multivariate data such as real microstructure images and real sintering signals: toward a SPS digital twin.

DOI:

https://doi.org/10.59499/EP246281353