AE-GCN-LSTM framework forecasts microstructure evolution in high-entropy alloys
July 31, 2026
This surrogate framework uses a multi-head autoencoder and Graph Convolutional Networks to predict phase evolution in AlCrFeNi alloys. It enables long-horizon forecasting up to 3,000,000 simulation timesteps without requiring retraining for unseen conditions.
HOW THIS AFFECTS YOU
●
researcherYou can bypass computationally expensive phase-field modeling for large-scale microstructure simulations.