Balanced optimization of multiple mechanical properties of homogeneous architecture hyperelastic material

Abstract

The fascinating mechanical properties of hyperelastic materials have been extensively studied. To meet the requirements of various application scenarios, researchers need to seek trade-offs and optimizations in different deformation modes, as these properties are often mutually restrictive. Machine learning with powerful nonlinear fitting ability, helps establish a balance between various mechanical properties, and facilitates iterative optimization in the manufacturing process of hyperelastic materials. Here, we propose a design strategy that reconciles the conflicting multiple mechanical properties of porous hyperelastic materials by using customized machine learning. Specifically, we combined multitask machine learning with targeted modules and domain knowledge from porous elastomer, and established the connection between the macroscopic structural parameters and multiple mechanical properties during the entire response process of hyperelastic materials obtained from additive manufacturing. By leveraging the connection, the contradiction between stiffness and energy dissipation in hyperelastic materials can be mitigated solely through macroscopic stacked structural optimization. The strategy is also employed to optimize the printing performance of silicone ink, demonstrating satisfactory results. Therefore, this strategy is expected to provide an efficient paradigm for simultaneously reconciling and optimizing the complex practical requirements of hyperelastic materials.