Real-time digital optimisation and decision making for energy and transport systems

能源和运输系统的实时数字优化和决策

• 批准号: EP/Y005619/1
• 负责人: Georgios Rigas
• 金额: $ 180.25万
• 依托单位: Imperial College London
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FY005619%2F1
• 关键词: Real; time; digital; optimisation; decision

In this project, we will seamlessly combine two disciplines that have been historically received continuous government and industrial funding: physics-based modelling, which is generalisable and robust but may require tremendous computational cost, and machine learning, which is adaptive and fast to be evaluated but not easily generalisable and robust. The intersection of the two spawns scientific machine learning, which maximises the strengths and minimises the weaknesses of the two approaches. The data will be provided by high-fidelity simulations and experiments, from the UK state-of-the-art facilities and software. The efficiency of the machine learning training will be maximised for the algorithms to require minimal energy (thereby, producing minimal emissions by minimising electricity consumption). This project builds upon large UK and EU funded expertise in scientific machine learning and simulation, which will be generalised to fast, real-time decision making. The most significant bottleneck of most scientific machine learning is that they need time to be re-trained offline when new data becomes available. We will transform offline paradigms into real-time approaches for the models to re-adapt and provide accurate estimates on the fly. This project will culminate into the delivery of practical digital twins (defined as digital counterparts of real world physical systems or processes that can be used for simulation, prediction of behaviour to inputs, monitoring, maintenance, planning and optimisation) to solve currently intractable problems in wind energy, hydrogen, and road transportation. This project will transfer the technical achievements and real-time digital twin to policy-making.

在这个项目中，我们将无缝地结合两个学科，这两个学科在历史上一直得到政府和工业的持续资助：基于物理的建模，它具有通用性和鲁棒性，但可能需要巨大的计算成本；以及机器学习，它具有适应性和快速评估性，但不容易通用性和鲁棒性。这两种方法的交叉产生了科学的机器学习，它最大限度地发挥了这两种方法的优势，并最大限度地减少了缺点。数据将由高保真模拟和实验提供，来自英国最先进的设施和软件。机器学习训练的效率将被最大化，因为算法需要最少的能量（因此，通过最小化电力消耗产生最少的排放）。该项目建立在英国和欧盟资助的大型科学机器学习和模拟专业知识的基础上，将推广到快速、实时的决策。大多数科学机器学习最重要的瓶颈是，当有新数据可用时，它们需要时间离线重新训练。我们将把离线范式转换为实时方法，以便模型重新适应并在飞行中提供准确的估计。该项目最终将交付实用的数字双胞胎（定义为现实世界物理系统或过程的数字对应，可用于模拟、预测输入行为、监测、维护、规划和优化），以解决目前在风能、氢和道路运输方面棘手的问题。该项目将把技术成果和实时数字孪生转化为决策。

项目成果

Inferring unknown unknowns: Regularized bias-aware ensemble Kalman filter

• DOI: 10.1016/j.cma.2023.116502
• 发表时间: 2023-06
• 期刊: Computer Methods in Applied Mechanics and Engineering
• 影响因子: 7.2
• 作者: Andrea N'ovoa;A. Racca;L. Magri

Active Flow Control for Bluff Body Drag Reduction Using Reinforcement Learning with Partial Measurements

• DOI: 10.1017/jfm.2024.69
• 发表时间: 2023-07
• 期刊: Journal of Fluid Mechanics
• 影响因子: 3.7
• 作者: C. Xia;Junjie Zhang;E. Kerrigan;Georgios Rigas