Fast Charging Batteries via Electrochemical Model-based Control

标题：基于电化学模型的控制快速充电电池在电信领域，2012年全球有52亿部活跃手机和超过1.7亿部手机销量。移动电话也是解决第三世界国家贫困和金融不平等的有力工具。在电气化交通方面，2012年美国销售了53,000辆插电式电动汽车。尽管销量不断增长，但里程焦虑被认为是电气化交通的最大阻碍。大大减少充电时间，例如相当于给油箱加油，将消除这一障碍，从而减少排放和对石油的依赖。很明显，快速充电增加了移动设备和电动汽车的实用性。然而，根据所使用的充电方法，它也会减少循环寿命。传统上，电池通过恒流/恒压（CCCV）协议充电。一般的手机从0-50%充电需要47分钟。然而，学术界和工业界都知道，替代协议可以缩短充电时间。然而，这些替代方案几乎总是启发式的，没有任何可证明的最优性质或安全约束满足保证。这个研究项目追求的是一种完全不同的、可能具有变革性的方法。本研究旨在通过发展电化学模型控制的控制理论基础，显著减少锂离子电池的充电时间。在数学上，这被表述为在与老化相关的估计电化学变量的约束下最小化充电时间。这种方法称为基于电化学模型的控制（ECC）。这种控制策略的发展将为电池在其电化学极限下的安全运行方式提供重大突破。PI目前正在建立合作，通过这项研究影响这两个部门。与劳伦斯伯克利国家实验室BATT小组的电化学家合作也在进行中。在加州大学伯克利分校，PI开设了一门名为“能源系统与控制”的新课程。本课程研究交通运输和能源基础设施中的能源系统，作为系统和控制理论的动机。本研究的结果将直接影响课程内容。最后，PI历来招收代表性不足的学生，以促进接受高等教育的机会。PI计划与加州大学伯克利分校STEM创新、领导力和多样性中心协调，招募本科生研究人员。最终，本研究寻求在500次充放电循环中实现10分钟0-20%的充电时间和25分钟0-50%的充电时间。该项目分为三个综合研究任务。(1)首先分析参数敏感性，建立面向快速充电的约简模型用于估计器设计。（ii）其次，推导出可证明的稳定状态估计器和最优快速充电算法。这些设计将分别推进PDE估计和参考调速器理论，同时将这些理论转化为电池系统。（iii）最后，它将量化ECC方法与-à-vis传统CCCV协议在电池在环实验设施上的性能。该项目是第一个专注于通过电化学模型优化电池充电时间的控制理论方法的项目。数学上，这些模型是多态耦合非线性偏微分方程（PDEs）。由于模型的复杂性，系统和控制理论中的几个基本工具将在电池的背景下发展。(i)第一个是评估多状态偏微分方程模型参数敏感性的系统程序。（ii）第二个是实现可观测性的与快速充电相关的模型缩减技术。（三）PDE-ODE模型状态估计理论的进展。（四）PDE-ODE模型参考调控理论的进展。(v)最后是对ECC快速充电性能与传统协议的实验量化。

Title: Fast Charging Batteries via ElectroChemical Model-Based Control In telecommunications, there were 5.2B active mobile handsets and over 1.7B mobile phone sales worldwide for 2012. Mobile phones are also a powerful tool for solving poverty and financial inequity in third world countries. In electrified transportation, there were 53,000 were plug-in electric vehicles sold in the U.S. for 2012. Despite growing sales, range anxiety is considered the largest inhibitor of electrified transportation. Significant reduction in charge times, e.g. comparable to filling a gas tank, would eliminate this obstacle and consequently reduce emissions and oil dependence. It is clear that fast charging increases the practicality of mobile devices and electric vehicles. However, it can also decrease cycle life depending on the charging method used. Traditionally, batteries are charged via a constant current/constant voltage (CCCV) protocol. A typical mobile phone requires 47 minutes to charge from 0-50%. However, it is well-known within the academic and industrial communities that alternative protocols can reduce charge times. Such alternatives, however, are almost always heuristic, without any provably optimal properties or safe constraint satisfaction guarantees. This research project pursues a drastically different and potentially transformative approach. This research seeks to significantly reduce Li-ion battery charge times by developing control theoretic foundations for electrochemical model-based control. Mathematically, this is formulated as minimizing charge time subject to constraints on estimated electro-chemical variables associated with aging. This approach is termed Electro-chemical model-based Control (ECC). The development of this control strategy will provide a major breakthrough in the way batteries are operated, safely at their electrochemical limits. The PI is currently establishing collaborations to impact both sectors with this research. Collaborations with electrochemists in the BATT group at Lawrence Berkeley National Lab are also being pursued. At UC Berkeley, the PI has created a new course entitled Energy Systems and Control. This course studies energy systems in transportation and energy infrastructures as motivation for systems and control theory. Results from this research will directly influence course material. Finally, the PI has historically recruited students of underrepresented populations to facilitate higher educational access. The PI plans to coordinate with the UC Berkeley Center for STEM Innovation, Leadership, and Diversity to recruit undergraduate researchers.Ultimately, this research seeks to achieve a 10 minute 0-20% charge time and 25 minute 0-50% charge time over 500 charge/discharge cycles. The project is organized into three integrated research tasks. (i) First, it will analyze parameter sensitivity and develop a fast charging-oriented reduced model for estimator design. (ii) Second, it will derive provably stable state estimators and optimal fast charging algorithms. These designs will advance PDE estimation and reference governor theory, respectively, while translating these theories to battery systems. (iii) Finally, it will quantify the performance of an ECC approach vis-à-vis traditional CCCV protocols on a battery-in-the-loop experimental facility. This project is among the first to focus on control-theoretic methods for optimizing battery charge times via electro-chemical models. Mathematically, these models are multi-state coupled nonlinear partial differential equations (PDEs). Due to the model complexity, several fundamental tools in systems and control theory will be developed in the context of batteries. (i) The first is a systematic procedure for assessing parameter sensitivity in multi-state PDE models. (ii) The second is fast charging-relevant model reduction techniques for achieving observability. (iii) The third is advancements to state estimation theory for PDE-ODE models. (iv)The fourth is advancements to reference governor theory for PDE-ODE models. (v) The last is an experimental quantification of ECC fast charging performance versus traditional protocols.