Modeling Fission Gas Release at the Mesoscale using Multiscale DenseNet Regression with Attention Mechanism and Inception Blocks
This work addresses the computational bottleneck in nuclear fuel modeling for researchers and engineers, though it is incremental as it applies existing deep learning techniques to a specific domain problem.
The authors tackled the problem of computationally intensive mesoscale simulations of fission gas release in nuclear fuel by developing a deep learning approach that predicts instantaneous FGR flux from 2D microstructure images, achieving high predictive power with R² values above 98% and a mean absolute percentage error of 4.4% for the best model.
Mesoscale simulations of fission gas release (FGR) in nuclear fuel provide a powerful tool for understanding how microstructure evolution impacts FGR, but they are computationally intensive. In this study, we present an alternate, data-driven approach, using deep learning to predict instantaneous FGR flux from 2D nuclear fuel microstructure images. Four convolutional neural network (CNN) architectures with multiscale regression are trained and evaluated on simulated FGR data generated using a hybrid phase field/cluster dynamics model. All four networks show high predictive power, with $R^{2}$ values above 98%. The best performing network combine a Convolutional Block Attention Module (CBAM) and InceptionNet mechanisms to provide superior accuracy (mean absolute percentage error of 4.4%), training stability, and robustness on very low instantaneous FGR flux values.