Mesh/membrane composite with superior mechanical performance: A deep learning-based design

The mesh/membrane composite structure in aerospace and engineering applications is often susceptible to degradation or failure due to lack of stiffness and strength. Here, we propose a method to design the mesh/membrane composite based on deep learning technique, trained with a database of thousands of structures from finite element simulations. Our method provides a programmable design solution to screen out desired candidates of mesh with superior fracture strength in specified relative density and porosity. Results show that the method can offer designs with different failure modes and crack regions for various target functionalities, while maintaining light weight and superior fracture strength. Building on the paradigm of smart design, we suggest potential strategies for designing the mesh/membrane composite with desired mechanical performance.