Physics-Based Probabilistic Prognostics for Battery Health Management

该项目的目标是为锂离子（Li-ion）电池创建一个基于物理的概率预测平台。锂离子电池技术在实现混合动力汽车、电动汽车和可再生能源的大规模采用方面发挥着越来越重要的作用。该平台的成功开发将在延长电池寿命和确保电池安全方面取得重大进展。能源储存管理方面的进步可以降低成本，促进混合动力、电动汽车和可再生能源的广泛采用，从而减少我国对外国能源的依赖。研究结果将通过与电池安全领域的行业领导者合作，传播给电池行业、主要利益相关者和决策者。一项包容性教育和推广计划将有助于(1)通过将研究成果纳入课堂教学，培养具有全球竞争力的电池可靠性建模劳动力；(2)为本科生和K-12学生提供研究经验，重点是增加女性和代表性不足的少数民族的参与；(3)通过在当地社区进行讲座，提高公众对电池可靠性和安全性的认识。所提出的预测平台的新颖之处在于它能够使用概率模型将机械退化分析集成到剩余使用寿命预测中。更具体地说，采用基于模型的平滑和学习，结合偏置校正半电池模型，从噪声电压和电流测量中推断退化程度。其次是基于物理的预测，其中基于模型的跟踪用于预测未来的退化轨迹和剩余使用寿命。由此产生的预测方法利用物理知识和传感器数据的组合来实现预测准确性和稳健性的提高。所提出的平台将通过进一步理解：(1)如何使用相对较少的昂贵实验验证机械半电池模型来推进电池健康管理领域；(2)如何仅利用早期数据预测长期退化；(3)物理在设计预测方法中的作用；(4)退化模式对容量衰减和失效的耦合效应。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The objective of this project is to create a physics-based probabilistic prognostics platform for lithium-ion (Li-ion) batteries. Li-ion battery technology is playing an increasingly important role in realizing wide-scale adoption of hybrid and electric vehicles, and renewable energy sources. Successful development of the proposed platform will produce major advancements in extending battery life while ensuring battery safety. Advances in energy storage management could reduce the costs and promote the wide-scale adoption of hybrid and electric vehicles and renewable energy sources, which in turn will reduce the dependence of our nation on foreign sources of energy. The research findings will be disseminated to the battery industry, main stakeholders, and decision makers through collaboration with an industry leader in battery safety. An inclusive education and outreach plan will help (1) train a globally competitive workforce in battery reliability modeling by incorporating research findings into classroom teaching, (2) provide research experiences to undergraduate and K-12 students with an emphasis on increasing the participation of women and underrepresented minorities, and (3) raise public awareness about battery reliability and safety by giving talks to the local communities. The novelty of the proposed prognostics platform is its ability to integrate mechanistic degradation analysis into remaining useful life prediction using probabilistic models. More specifically, model-based smoothing and learning are adopted, in conjunction with bias-corrected half-cell models, to infer the degrees of degradation from noisy voltage and current measurements. This is followed by physics-based prognostics, where model-based tracking is used to predict future degradation trajectories and remaining useful life. The resulting prognostics approach leverages a combination of physical knowledge and sensor data to achieve gains in prediction accuracy and robustness. The proposed platform will advance the field of battery health management by furthering understanding on: (1) how to validate mechanistic half-cell models using relatively few expensive experiments; (2) how to predict the long-term degradation using only early-life data; (3) the role of physics in designing prognostics approaches; and (4) the coupling effects of degradation modes on capacity fade and failure.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.