Collaborative Research: Inferring The In Situ Micro-Mechanics of Embedded Fiber Networks by Leveraging Limited Imaging Data

合作研究：利用有限的成像数据推断嵌入式光纤网络的原位微观力学

• 批准号: 2127864
• 负责人: Emma Lejeune
• 金额: $ 28.69万
• 依托单位: Trustees of Boston University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-01-01 至 2024-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2127864&HistoricalAwards=false
• 关键词: Collaborative; Research; Inferring; Situ; Micro

This grant will focus on gaining a fundamental understanding of embedded fiber networks and creating the tools necessary to characterize their behavior from limited available measurements. Embedded fiber networks are ubiquitous in nature, from the extracellular matrix surrounding biological cells, to branching blood vessels embedded in organs, to moth’s cocoons. Understanding these systems is important because these systems are the fundamental mechanical building blocks of many types of natural and engineered biological tissue, and bio-inspired advanced materials. It is important not only to understand these systems, but also to be able to measure their mechanical behavior in a non-destructive manner so that advances in understanding can be applied in the real world. This research project will synthesize experiments, theory-based computational models, and data-driven computational models to elucidate the fundamental relationship between embedding matrix properties, fiber properties, and fiber network properties for soft embedded fiber networks undergoing large deformation. In addition, this research project will develop computational capabilities for the analysis of these systems where severely limited image-based data is used to predict both structural properties and characterize mechanical behavior. The research will be complemented by disseminating relevant data and code under open source licenses, and releasing online modules focused on applying machine learning to mechanics research. The research will also be complemented by establishing educational outreach programs at the middle school and high school levels that focus on bringing STEM education to underserved populations. The specific goal of this research is to define fundamental structure-function relationships in soft embedded fiber networks undergoing large deformation and create the tools needed to analyze these systems given limited available imaging data. Critically, it is necessary to develop tools to evaluate these systems non-destructively because one of their most important applications is in living systems. Thus, the research objectives of this project include: (i) curating an experimental dataset and implementing and validating a computational model of three-dimensional embedded fiber networks undergoing large deformation; (ii) understanding and delineating the different mechanical regimes of embedded fiber networks undergoing large deformation; (iii) establishing and testing a machine learning framework to rapidly and non-destructively analyze embedded fiber networks from imperfectly-paired images taken on the discrete fiber scale. The project will allow the PIs to advance the knowledge base at the interface of applied mechanics, computational mechanics, and machine learning, and establish their long-term careers in the mechanics of materials and structures.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

这笔赠款将重点用于获得对嵌入式光纤网络的基本了解，并创建必要的工具来通过有限的可用测量来表征其行为。嵌入式光纤网络在自然界中无处不在，从生物细胞周围的细胞外基质到嵌入器官中的分支血管。 ，对于蛾子的茧来说，了解这些系统很重要，因为这些系统是许多类型的天然和工程生物组织以及仿生先进材料的基本机械构件。不仅了解这些系统很重要，而且了解这些系统也很重要。能够以非破坏性的方式测量它们的机械行为，以便将理解的进步应用于现实世界。该研究项目将综合实验、基于理论的计算模型和数据驱动的计算模型，以阐明嵌入矩阵之间的基本关系。此外，该研究项目将开发用于分析这些系统的计算能力，其中使用严格有限的基于图像的数据来预测结构特性和表征。研究将是机械行为。通过在开源许可下传播相关数据和代码，并发布专注于将机器学习应用于力学研究的在线模块，该研究还将通过在初中和高中层面建立专注于 STEM 教育的教育推广计划来补充。这项研究的具体目标是定义经历大变形的软嵌入光纤网络中的基本结构-功能关系，并在有限的可用成像数据的情况下创建分析这些系统所需的工具。非破坏性地评估这些系统因为它们最重要的应用之一是在生命系统中，因此，该项目的研究目标包括：（i）整理实验数据集并实现和验证经历大变形的三维嵌入式光纤网络的计算模型；了解和描绘经历大变形的嵌入式光纤网络的不同机械状态；（iii）建立和测试机器学习框架，以根据离散光纤尺度上拍摄的不完美配对图像快速、无损地分析嵌入式光纤网络。允许PI 致力于推进应用力学、计算力学和机器学习领域的知识基础，并在该奖项中建立其长期的力学职业生涯，体现了材料和结构。该奖项反映了 NSF 的法定使命，并被认为值得支持通过使用基金会的智力优点和更广泛的影响审查标准进行评估。

项目成果

Learning mechanically driven emergent behavior with message passing neural networks

使用消息传递神经网络学习机械驱动的紧急行为

• DOI: 10.1016/j.compstruc.2022.106825
• 发表时间: 2022-10
• 期刊: Computers & Structures
• 影响因子: 4.7
• 作者: Prachaseree, Peerasait;Lejeune, Emma
• 通讯作者: Lejeune, Emma

Investigating deep learning model calibration for classification problems in mechanics

研究力学分类问题的深度学习模型校准

• DOI: 10.1016/j.mechmat.2023.104749
• 发表时间: 2023-09
• 期刊: Mechanics of Materials
• 影响因子: 3.9
• 作者: Mohammadzadeh, Saeed;Prachaseree, Peerasait;Lejeune, Emma
• 通讯作者: Lejeune, Emma

Locality sensitive hashing via mechanical behavior

通过机械行为进行局部敏感哈希

• DOI: 10.1016/j.eml.2023.102042
• 发表时间: 2023-03-09
• 期刊: Extreme Mechanics Letters
• 影响因子: 4.7
• 作者: E. Lejeune;Peerasait Prachaseree
• 通讯作者: Peerasait Prachaseree