CAREER: Reconfigurable and Predictive Control with Reinforcement Learning Supervisor for Active Battery Cell Balancing

该项目将支持为电池电动汽车提供新方法的研究，通过实现更环保的电动汽车，促进科学进步，促进国家健康和繁荣。由于电池的变化，电动汽车中有大量的电力未被使用，这引起了人们对效率和可持续性的担忧。有源电池平衡控制旨在解决这些问题，但由于许多汽车系统固有的实时限制和有限的传感能力，现有方法在纯电动汽车上的成功有限。该项目支持解决大规模系统控制、控制融合、智能控制和电池管理等主要挑战的基础研究。新的设计和方法将为主动电池平衡控制提供一个变革性的框架，将车辆水平信息无缝集成到电池管理中，以提高电动汽车的能源效率和行驶里程。这项研究与开发有效的交通系统以应对气候变化相关的关键社会目标具有协同作用。因此，这项研究的结果将有利于美国的经济、生活质量和健康。这项研究涉及多个学科，包括控制理论、强化学习、传感和估计以及电池管理。多学科方法还促进了代表性不足的群体参与研究，并对工程教育和汽车劳动力产生积极影响。主动电池平衡控制有望大大提高电池电动汽车的效率，增加续驶里程，并提高公众接受度。为了实现这一目标，我们计划实现四个紧密结合的研究目标：1)开发一个有效的电池水平估计框架，在有限的测量下估计电池的充电状态和容量，而不需要大量的计算；2)开发一种新的预测单元平衡控制框架，将车速预览与快速鲁棒的预览误差自适应无缝集成；3)开发可重构控制框架，在不降低整体平衡控制性能的前提下，扩展电动汽车电池单体平衡控制算法；4)通过广泛的模拟和实验来评估和验证所提出的框架。总的来说，这些研究工作的进展有望使电动汽车更容易被接受，更便宜，并将为大规模动力系统创造新的计算效率控制机制。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。

This project will support research that will contribute novel methodologies related to battery electric vehicles, promoting the progress of science and advancing national health and prosperity by enabling greener electric vehicles. Due to battery cell variations, significant amount of electricity in electric vehicle remain unused, which raises concerns on efficiency and sustainability. Active battery cell balancing control aims to address these concerns, but existing methods have limited success in battery electric vehicles due to concerns in real-time constraints and limited sensing capability inherent to many automotive systems. This project supports fundamental research that addresses the major challenges in control of large-scale systems, control fusion, intelligent control, and battery management. The new designs and methodologies will offer a transformative framework in active battery cell balancing control that seamlessly integrate vehicle level information into battery management to improve the energy efficiency and driving range of electric vehicles. This research is synergistic with key societal goals related to developing efficient transportation systems for combating the climate change. Therefore, results from this research will benefit the U.S. economy, life quality, and health. This research involves several disciplines including control theory, reinforcement learning, sensing and estimation, and battery management. The multi-disciplinary approach also facilitates the participation of underrepresented groups in research and positively impacts engineering education and automotive workforce. The active battery cell balancing control is expected to greatly enhance the efficiency of battery electric vehicles, increase the driving range, and improve public acceptance. In pursuit of this goal, four closely integrated research objectives are planned: 1) Develop an efficient cell level estimation framework to estimate cell state-of-charge and capacity under limited measurements without incurring heavy computation; 2) Develop a novel predictive cell balancing control framework to seamlessly integrate vehicle speed preview with fast and robust adaptation to preview errors; 3) Develop a reconfigurable control framework to scale up cell balancing control algorithm for EV batteries without degrading overall balancing control performance; and 4) Evaluate and validate the proposed framework through extensive simulations and experiments. Collectively, advances from these research endeavors are expected to make electric vehicles more acceptable and more affordable, and it will create new computationally-efficiency control mechanism for large scale dynamical systems.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.