The digital twin of fusion energy components

主要目标/宗旨：（1）开发一个复制物理实验的数字孪生平台：该项目将探索为聚变能应用建立数字孪生的可能性。这包括纯数据驱动的数字孪生和基于物理的数字孪生平台。该项目的这一部分包括对现有文献的深入调查，人工智能培训，数据收集和计算力学。(2)使用逆方法论和人工智能构建物理学知识的数字双胞胎：构建数字双胞胎的重大挑战之一是一个可以用稀疏数据输入复制现实的平台。因此，本部分将使用人工智能、物理学神经网络和计算力学进行逆向分析，以了解数据缺口是否可以稳健和准确地填补。此外，这一步将研究从稀疏数据中建立完整问题的可能性。(3)将（1）和（3）放在一起，根据测试数据验证数字孪生模型：将采用步骤1和2中的经验教训来研究组件测试的复制精度。这一步涉及到实验的回顾性复制和预测系统/组件的未来行为。新奇：结合数据、人工智能和计算力学来提供基于物理的数字孪生模型是一个新概念。使用逆模型、稀疏数据、人工智能和计算力学来建立整个问题是物理科学和工程挑战。

Key objectives/Aims: (1) Developing a digital twin platform that replicates physical experiments: The project will explore the possibilities of building a digital twin for fusion energy applications. This includes purely data-driven digital twin and physics-based digital twin platforms. This part of the project consists of an intensive survey of available literature, training in AI, data collection, and computational mechanics. (2) Using the inverse methodology and AI to build a physics-informed digital twin: One of the significant challenges in building a digital twin is a platform that can replicate the reality with sparse data input. Therefore, this part will use inverse analysis using AI, physics informed neural networks, and computational mechanics to understand whether the data gap can be robustly and accurately filled. In addition, this step will investigate the possibility of establishing a problem in its entirety from sparse data. (3) Putting (1) and (3) together to verify the digital twin against test data: The lessons learnt in steps 1 and 2 will be employed to investigate how accurately component testing may be replicated. This step involves retrospective replication of experiments and forecasting the future of how the system/components behave. Novelty: Combining data, AI and computational mechanics to deliver a physics-based digital twin is a new concept. Establishing the problem in its entirety using inverse models, sparse data, AI, and computational mechanics is the physical sciences and engineering challenge.