GrainNN: A neighbor-aware long short-term memory network for predicting microstructure evolution during polycrystalline grain formation
GrainNN: A neighbor-aware long short-term memory network for predicting microstructure evolution during polycrystalline grain formation
复制标题
DOI：
10.1016/j.commatsci.2022.111927
复制
发表时间：
2023-02
影响因子：
3.3
通讯作者：
Yigong Qin;S. DeWitt;B. Radhakrishnan;G. Biros
中科院分区：
文献类型：
--
作者：
Yigong Qin;S. DeWitt;B. Radhakrishnan;G. Biros
高保真模拟合金中的晶粒形成是表征工艺-机械性能的不可或缺的工具。然而，这样的模拟可能计算昂贵，因为它们需要精细的空间和时间离散。它们的成本成为参数研究和集成运行的障碍，并最终使最优控制和不确定性量化等下游任务具有挑战性。为了实现这种下游任务，我们引入了GrainNN，这是一种高效而准确的降阶模型，用于在添加剂制造条件下进行外延晶粒生长。GrainNN是一种序列到序列的长期短期记忆(LSTM)深度神经网络，它进化了手动制作的特征的动力学。它的创新是(1)具有颗粒微结构特定变压器架构的注意机制；以及(2)网络的几个克隆的重叠组合，以概括为不同于用于训练的颗粒配置。这种设计使GrainNN能够为看不见的物理参数、颗粒数量、区域大小和几何形状预测颗粒形成。此外，GrainNN不仅可以重建感兴趣的量，而且可以逐点准确地进行。在数值实验中，我们使用多晶相场方法来生成训练数据和评估GrainNN。对于多参数、多颗粒的集成模拟，GrainNN可以比相场模拟快几个数量级，同时提供5%-15%的逐点误差。该加速比包括用于生成训练数据的相场模拟的成本。
High fidelity simulations of grain formation in alloys are an indispensable tool for process-to-mechanical-properties characterization. Such simulations, however, can be computationally expensive as they require fine spatial and temporal discretizations. Their cost becomes an obstacle to parametric studies and ensemble runs and ultimately makes downstream tasks like optimal control and uncertainty quantification challenging. To enable such downstream tasks, we introduce GrainNN, an efficient and accurate reduced-order model for epitaxial grain growth in additive manufacturing conditions. GrainNN is a sequence-to-sequence long-short-term-memory (LSTM) deep neural network that evolves the dynamics of manually crafted features. Its innovations are (1) an attention mechanism with grain-microstructure-specific transformer architecture; and (2) an overlapping combination of several clones of the network to generalize to grain configurations that are different from those used for training. This design enables GrainNN to predict grain formation for unseen physical parameters, grain number, domain size and geometry. Furthermore, GrainNN not only reconstructs the quantities of interest but also can be pointwise accurate. In our numerical experiments, we use a polycrystalline phase field method to both generate the training data and assess GrainNN. For multiparametric, ensemble simulations with many grains, GrainNN can be orders of magnitude faster than phase field simulations, while delivering 5%–15% pointwise error. This speedup includes the cost of the phase field simulations for generating training data.