Microstructure-Conditioned Surrogate Models for Graded Multiscale Optimization of Mycelium Composites
The research developed a conditioned surrogate model using a hypernetwork to accurately predict mechanical behavior for mycelium-woodchip composites, even with limited data. This allowed for tractable multiscale simulations and optimization of functionally graded structures. By conditioning the model on both microstructural variables and manufacturing inputs, the method successfully optimized a graded disk, reducing peak stress by 43% compared to random microstructures.