Collaborative Research: Multi-Accuracy Bayesian Models for Improving Property Prediction of Nanotube Buckypaper Composites

佛罗里达州立大学和德克萨斯农工大学之间的这项合作研究旨在开发多精度预测模型，以提高对纸基复合材料性能的预测能力。该团队将研究适当的建模策略，以整合多精度信息，以及解决相关计算和设计问题的解决技术，以保证方法的效率和实用性。目前，用于预测厚纸基复合材料性能的力学模型不多，但由于模型的不完备和不确定性，大多数模型精度较低。力学模型和实际物理实验的输出构成了一套多精度信息源，从不同的角度反映了相同的物理性质。我们的猜想是，结合多精度输出可以帮助增强基于厚纸的复合材料的预期性能预测。这种新方法的成功开发将有可能为厚纸基复合材料提供稳定、可重复和可扩展的生产工艺。厚纸基复合材料是最受欢迎的纳米材料之一，因为它具有传统材料所没有的特性，并且适用于广泛的应用。首席PI隶属于佛罗里达州立大学的高性能材料研究所（HPMI），在纸的研发和原型生产能力方面是美国最好的研究机构之一。预测模型是实现纳米制造过程和质量控制的基石，因为只有使用这些模型，人们才能识别关键的过程变量，以便进行过程测量或进行调整，以产生预期的结果。

This collaborative research between Florida State University and Texas A&M University is to develop multi-accuracy predictive models that can enhance the prediction capability for bukypaper-based composite properties. The team will investigate proper modeling strategies to integrate the multi-accuracy information as well as the solution techniques that address the associated computational and design issues in order to guarantee the method's efficiency and practicality. Currently, a few mechanics models are available for making property predictions for bukypaper-based composites but most of them suffer from having low accuracy due to model inadequacy and uncertainty. The outputs from the mechanics models and actual physical experiments constitute a set of multi-accuracy information sources, reflecting the same physical properties from different perspectives. Our conjecture is that combining the multi-accuracy outputs could help enhance the desired property predication for bukypaper-based composites.The successful development of this new methodology will potentially enable stable, repeatable, and scalable production processes for bukypaper-based composites, which are one of the most sought-after nano-materials, due to its properties unfound in traditional materials and applicability to a broad array of applications. The High-Performance Material Institute (HPMI) at Florida State University, with which the lead PI is affiliated, is one of the best research facilities in the nation in terms of buckypaper R&D and prototype production capabilities. Predictive models are the cornerstones for enabling any attempts of process and quality control in nano-manufacturing because only with these models can people identify the critical process variables for taking in-process measurements, or making adjustments, in order to yield expected outcomes.