CAREER: Universal Modeling of Real and Virtual Energy Storage with Connected Power Electronics

现代高度动态的电力系统，如可再生能源主导的电网，电动汽车和飞机，需要储能来提供弹性，以更小的时间尺度减轻功率变化。存在多种“真实的”能量存储物理，例如电池、超级电容器、燃料电池，但是在能量或功率带宽方面受到限制。最近，调制负载控制，特别是通过灵活的热负载，如建筑物HVAC系统或热水器，已获得理解，并被视为“虚拟”存储，鉴于其固有的存在，作为一个能量缓冲器周围的标称负载。储能系统的典型设计考虑一个或多个存储物理，并聚合它们各自的电气模型用于控制和能量管理;这种方式忽略了系统级的画面，并在创建后处理复杂性。本研究探索一种自上而下的协同设计过程，利用通用的储能电气模型，结合真实的和虚拟的储能和连接的电力电子。所提出的过程直接在一组能量存储选项中产生最佳的能量存储要求，并允许最大的弹性，同时减少复杂的能量管理负担并维持延长的寿命。这项工作将使交通电气化，公用事业和建筑部门的许多利益相关者受益。此外，该项目还将通过结合电力电子技术与储能系统的进步来加强本科生和研究生课程。该项目将与管理咨询和国际教育合作伙伴合作，培养STEM学生的领导力和全球视野。此外，该项目还将面向当地的小企业和美洲原住民部落，提供新兴储能技术方面的教育，从而使人们能够在该领域追求繁荣的职业生涯。该项目为储能系统开发了一种新颖的通用建模和设计框架。传统的能量存储的启发式选择是次优的，真实的和虚拟存储的动态交互是缺乏的，虚拟能量存储的电力电子方面没有得到很好的理解，并且组合存储的寿命影响是未知的。建模和设计的基础研究强调使用增强的多时标等效阻抗网络进行带宽选择，然后是连接电力电子设备的协同设计，多目标优化，动态能量管理和可靠性分析。智能搜索、深度强化学习和大数据驱动控制中的几种基于机器学习的方法也将在工作中发挥作用。科学知识将通过深入的电-热-机械物理、现代电力电子转换器和控制、新兴的数据科学和人工智能以及硬件测试平台获得。该奖项反映了NSF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。

Energy storage supports the uninterruptable operation of electricity, mechanics, heat, etc. Modern highly-dynamic electrical systems, such as renewable-dominated grids, electric vehicles and aircraft, require energy storage to provide resilience, mitigating power variations at much smaller timescales. Multiple “real” energy storage physics exist, such as batteries, supercapacitors, fuel cells, but are limited in energy or power bandwidth. Recently, modulated load control, particularly through flexible thermal loads, such as building HVAC systems or water heaters, has gained understanding and is regarded as “virtual” storage, given its inherent existence as an energy buffer around a nominal load. A typical design of an energy storage system considers one or multiple storage physics and aggregates their respective electric models for control and energy management; this way overlooks the system-level picture and handles the complexity after it is created. This research explores a top-down co-design process by utilizing a universal energy storage electric model, combining real and virtual energy storage and connected power electronics. The proposed process directly yields optimal energy storage requirements among a cluster of energy storage options, and allows for most resiliency while reducing the complex energy management burden and sustaining an extended lifetime. The work will benefit many stakeholders in transportation electrification, utilities, and building sectors. Further, the project will enhance undergraduate and graduate curriculum by combining the advances in power electronics tied energy storage systems. The project will work with management consulting and international educational partners to train STEM students toward leadership and global vision. In addition, the project will reach out to local small businesses and Native American tribes to provide education in emerging energy storage technologies, thereby enabling pursuit of prosperous careers in the field.This project develops a novel universal modeling and design framework for energy storage systems. Conventional heuristic selection of energy storage is suboptimal, dynamic interaction of real and virtual storage is lacking, the power electronics side of virtual energy storage is not well understood, and the lifetime impact of combined storage is not known. Fundamental research in modeling and design emphasizes bandwidth selections using an enhanced multi-timescale equivalent impedance network, followed by a co-design of connected power electronics, multi-objective optimization, dynamic energy management, and reliability analysis. Several machine learning-based methods in intelligent search, deep reinforcement learning, and big-data-driven control will also play a role in the work. Scientific knowledge will be gained through in-depth electrical-thermal-mechanical physics, modern power electronic converters and controls, emerging data science and artificial intelligence, and a hardware testbed.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.