Towards Safer and Resilient Batteries via Active Diagnostics and Fault-tolerant Control

最近发生的波音787梦幻客机、特斯拉型号S和三星Galaxy Note 7的电池故障事件强烈强调了需要更智能的电池安全解决方案。该项目旨在通过开发一种新的实时诊断框架来提高电池的安全性。提出的框架旨在通过设计实时算法来诊断和控制各种电池故障，从而提供电池安全解决方案。除了提高电动飞机、医疗设备和消费电子等电池供电系统的安全性外，更安全的电池还将加快电动汽车和可再生能源系统的普及，从而带来环境效益、能源安全和经济稳定。该项目的结果还将促进对实时电池故障的理解，为电池科学家/设计师创造更好的电池提供见解。为该项目开发的理论控制/诊断工具可以很容易地扩展到其他多物理系统，如燃料电池、建筑和微型机电系统。结果将通过首席研究员的课程和出版物传播，增强汽车/能源部门下一代技术劳动力的知识。最后，该项目将与国家可再生能源实验室、非博士授予机构(如丹佛社区学院)和代表性不足的学生创造合作/指导机会。该项目的目标是通过设计基于多物理模型的实时算法来开发电池安全解决方案，这些算法(I)在早期阶段诊断电化学、机械、电气和热故障，以及(Ii)采取纠正控制措施，将故障影响/传播降至最低。这些算法除了最大限度地提高能量利用率、循环寿命和可靠性外，还将增加电池管理系统的容错能力。从实时诊断和控制的角度来看，这个项目将解决电池中的两个关键问题：缺乏测量(弱可观测性)和有限的驱动。将执行以下研究任务：(I)基于多物理的电池故障建模；(Ii)确定故障诊断能力的分析；(Iii)从建模和测量的不确定性中区分故障；(Iv)故障的实时检测、隔离和估计；以及(V)故障传播的实时控制。这些研究任务将产生以下工具：(I)用于确定故障诊断能力的离线分析工具，以及(Ii)用于故障诊断和容错控制的在线算法。这些工具/算法将利用分散和相互关联的系统理论、多目标最优控制理论和数据驱动的不确定性建模技术相结合来开发。算法的性能将根据真实世界的电池故障数据进行验证和量化。这一奖项反映了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。

Recent battery failure incidents of Boeing 787 Dreamliner, Tesla Model S and Samsung Galaxy Note 7 strongly emphasize the need for smarter battery safety solutions. This project aims to enhance battery safety by developing a novel real-time diagnostics framework. The proposed framework intends to provide battery safety solutions by designing real-time algorithms that diagnose and control various battery faults. Besides enhancing the safety of battery-powered systems such as electric aircrafts, medical equipment, and consumer electronics, safer batteries will accelerate penetration of electrified vehicles and renewable energy systems, leading to environmental benefits, energy security, and economic stability. The results of this project will also advance the understanding of real-time battery failures, providing insights to the battery scientists/designers to create better batteries. The theoretical control/diagnostics tools developed for the project can be readily extended to other multi-physics systems such as fuel cells, buildings, and micro electro-mechanical systems. The results will be disseminated through principal investigator's course and publications, enhancing the knowledge of next generation technological workforce in automotive/energy sectors. Finally, this project will create collaborative/mentoring opportunities with the National Renewable Energy Laboratory, non-PhD granting institutions such as the Community College of Denver, and underrepresented student The goal of this project is to develop battery safety solutions by designing multi-physics model-based real-time algorithms that (i) diagnose electrochemical, mechanical, electrical and thermal faults at early stages, and (ii) take corrective control action to minimize the fault effect/propagation. These algorithms will add fault-tolerance to battery management systems, besides maximizing energy utilization, cycle life and reliability. From real-time diagnosis and control viewpoint, this project will address two key issues in batteries: lack of measurements (weak observability) and limited actuation. The following research tasks will be performed: (i) multi-physics-based battery fault modeling, (ii) analysis to determine ability to diagnose faults, (iii) distinguishing faults from modeling and measurement uncertainties, (iv) real-time detection, isolation and estimation of faults, and (v) real-time control of fault propagation. These research tasks will produce the following tools: (i) offline tools for analysis to determine ability to diagnose faults, and (ii) online algorithms for fault diagnostics and fault-tolerant control. Such tools/algorithms will be developed utilizing a combination of decentralized and interconnected systems theory, multi-objective optimal control theory, and data-driven uncertainty modeling techniques. The performance of the algorithms will be validated and quantified on real-world battery failure data.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.