RAPID: Collaborative Research: Modeling, Analysis and Control of COVID-19 Spread in an Aircraft Cabin using Physics Informed Deep Learning

该项目将通过机舱内的空气传播感染来模拟、分析、预测和展示新冠肺炎暴发的控制机制。随着航空旅行的恢复，预计许多乘客会接触到新冠肺炎病毒，甚至可能被感染。因此，迫切需要迅速开发解决方案，通过了解飞机内部气流的动力学来确定传染的速度。这项研究将结合代表流体动力学、标量传输、流行病学和空气传播的四个独立的多物理模型来分析新冠肺炎在封闭系统(如飞机)内的传播。这项研究的多学科性质将在计算数学、深度学习、数据科学、流行病学和流体动力学的界面上产生新的算法，并将提供可直接应用于大规模数据的新技术，以实现高效和强大的数据分析。该项目还将作为对学生的宝贵培训。开源代码将向用户社区提供，并将向最终用户、学术研究人员、行业成员、从业者和政府研究实验室开放。这项研究还将扩展到其他物理空间，如海洋船舶、火车、公交车或任何其他公共交通工具。本研究将完成以下具体目标：(A)开发一个捕捉真实几何并耦合四个不同物理和生物系统的全三维计算模型；(B)实现一个隐藏的多物理神经网络框架，以实现数据同化；(C)除了开发和研究新的控制和强化学习机制外，还将使用模拟、实验和观测数据来评估预测能力。该框架考虑了可能没有戴口罩的飞机上受新冠肺炎感染的成员呼出水滴的特征，跟踪这些水滴的扩散，并通过这些耦合的多物理模型跟踪易感乘客吸入水滴的过程。这项研究将有助于开发一种新的物理信息深度学习框架，该框架将能够对建模为神经网络的多物理方程系统进行编码，同时与几何或初始和边界条件无关。千年发展目标的进展将提供数据驱动发现方面的进展，这将使人们能够更好地理解新冠肺炎的影响。这笔赠款是使用分配给议员的冠状病毒援助、救济和经济安全(CARE)法案附录提供的资金授予的。这一奖励反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This project will model, analyze, predict, and present control mechanisms for a COVID-19 outbreak through an airborne infection in an aircraft cabin. As air travel resumes, it is expected that many passengers would be exposed to and possibly infected by the COVID-19 virus. As a result, there is an urgent need to rapidly develop solutions to determine the speed of the contagion by understanding the dynamics of the airflow inside aircraft. This research will combine four separate multi-physics models representing fluid dynamics, scalar transport, epidemiology, and airborne infection to analyze the spread of COVID-19 within a closed system such as an airplane. The multidisciplinary nature of this research will yield new algorithms at the interface of computational mathematics, deep learning, data science, epidemiology, and fluid dynamics and will provide novel techniques that can be directly applied to large-scale data to allow efficient and powerful data analysis. The project will also serve as valuable training for students. Open-source codes will be made available to the user community and will be open to contributions from end-users, academic researchers, industry members, practitioners, and government research labs. The research may also be extended to other physical spaces, such as marine vessels, trains, buses, or any other medium of public transportation systems.This research will accomplish the following specific objectives (a) develop a fully 3-dimensional computational model capturing realistic geometry and coupling four different physical and biological systems; (b) implement a hidden multi-physics neural network framework to enable data assimilation and; (c) evaluate the predictive capability using simulated, experimental and observational data in addition to developing and studying novel control and reinforcement learning mechanisms. The framework considers the characteristics of the exhalation of the droplets from COVID-19 infected members on an airplane that may not be wearing face masks, tracking the dispersion of these droplets, and tracking the inhalation of the droplets by susceptible passengers through these coupled multi-physics models. The research will help to develop a novel physics-informed deep-learning framework that will be capable of encoding the multi-physics system of equations modeled into the neural networks while being agnostic to the geometry or the initial and boundary conditions. Progress on the goals will provide advances in data-driven discovery, which will allow a better understanding of the impact of COVID-19. This grant is being awarded using funds made available by the Coronavirus Aid, Relief, and Economic Security (CARES) Act supplement allocated to MPS.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.