Data-driven investigation of three-dimensional instabilities in magneto-active thin films heterogeneously patterned by design

我们建议调查和利用实验，理论，数值和数据集成分析，固有的不稳定性存在于磁流变弹性体（MRE）为基础的材料和结构。MRE是智能材料，其包括嵌入在聚合物基质中的可磁化颗粒，所述可磁化颗粒可以由磁场致动。在早期的工作之后，我们计划通过在位于被动或MRE基底上的MRE膜上赋予非均匀的磁机械特性来创建和利用可逆的按需远程可控的3D表面图案。我们的目标是实现这一目标，利用液体增材制造，并在并行，通过开发理论和计算工具来分析后分叉MRE膜模式下耦合磁-机械负载。实验，理论和计算信息将使用机器学习以数据驱动的方法进行合成，以便快速预测不稳定性以及解决逆问题。这些发展对于我们通过机器学习设计目标模式的最终目标至关重要，这可以显着减轻数值设计研究的计算负担。

We propose to investigate and exploit experimentally, theoretically, numerically and by data-integrated analysis, the instabilities inherently present in magnetorheological elastomer (MRE)-based materials and structures. MREs are smart materials comprising magnetizable particles embedded in a polymeric matrix that can be actuated by a magnetic field. Following earlier works, we plan to create and harness reversible on-demand remotely-controllable 3D surface patterns by imparting non-uniform magneto-mechanical properties on an MRE film resting on a passive or MRE substrate. We aim at achieving this by exploiting liquid additive manufacturing and, in parallel, by developing theoretical and computational tools to analyze the post-bifurcated MRE film patterns under coupled magneto-mechanical loading. The experimental, theoretical and computational information will be synthesized in a data-driven approach using machine learning in order to allow rapid forward predictions of the instabilities as well as for solving the inverse problem. Such developments are crucial for our ultimate goal of designing targeted patterns via machine learning, which could remarkably alleviate the computational burden of numerical design studies.