Artificial Intelligence Based Assessment of LiIon Battery Failure

基于人工智能的锂离子电池故障评估

• 批准号: 522646-2017
• 负责人: Fowler, Michael
• 金额: $ 1.82万
• 依托单位: University of Waterloo
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=640030
• 关键词: Artificial; Intelligence; Based; Assessment; LiIon

This Engage project aims to establish a new collaboration between Acerta Analytics Solutions and the researchteam supported by Dr. Michael Fowler. The two parties have not engaged before in any form of collaborationto date. Advances in computing technology enable new forms of data analysis. Acerta offers analyticssolutions to identify warranty issues of manufactured goods at the end of life. This means that Acerta'salgorithms analyze data collected from the manufactured good before it leaves the factory and based on the datadetermines whether the good should be shipped to the customer or whether it will likely have a warranty issueand thus should be inspected in the factory prior to shipping. Currently, Acerta is applied to products usinggearboxes, engines, and bearings in vehicles. Acerta is extending its product line to also offer analytics forbatteries. To extend the product line to batteries, Acerta needs to tune its algorithms and products to providemeaningful estimates of the quality of a battery system at the end of life. This requires answering the followingquestions (among others): (1) what is the state of the charge, (2) what is the health of the cells, and (3) what isthe expected state of degradation over the next years. Since Acerta uses machine learning mixed with domainknowledge, it needs domain expertise on feature engineering (i.e., determining what sensor values should beprovided) and feature processing (i.e., use features to extract information gain).

该Engage项目旨在建立Acerta Analytics Solutions和Michael Fowler博士支持的研究团队之间的新合作。到目前为止，双方没有进行过任何形式的合作。计算机技术的进步使新形式的数据分析成为可能。Acerta提供分析解决方案，以确定制成品在生命周期结束时的保修问题。这意味着Acerta的算法在成品出厂前分析收集到的数据，并根据这些数据确定产品是否应该发货给客户，或者是否可能有保修问题，因此应该在发货前在工厂进行检查。目前，Acerta应用于车辆中使用齿轮箱，发动机和轴承的产品。Acerta正在扩展其产品线，以提供电池分析。为了将产品线扩展到电池，Acerta需要调整其算法和产品，以便在电池系统寿命结束时提供有意义的质量估计。这需要回答以下问题（其中包括）：(1)电荷的状态是什么，(2)细胞的健康状况如何，(3)未来几年的预期降解状态是什么。由于Acerta将机器学习与领域知识混合使用，因此它需要在特征工程（即确定应该提供哪些传感器值）和特征处理（即使用特征提取信息增益）方面的领域专业知识。