Collaborative Research: CISE-MSI: DP: IIS RI: Research Capacity Expansion via Development of AI Based Algorithms for Optimal Management of Electric Vehicle Transactions with Grid

合作研究：CISE-MSI：DP：IIS RI：通过开发基于人工智能的算法来扩展研究能力，以实现电动汽车与电网交易的优化管理

• 批准号: 2318612
• 负责人: Ganesh Venayagamoorthy
• 金额: $ 30万
• 依托单位: Clemson University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-10-01 至 2026-09-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2318612&HistoricalAwards=false
• 关键词: Collaborative; Research; CISE; MSI; DP

Pressing challenges such as climate change and the necessity to reduce carbon emissions require the transition from gasoline-powered vehicles to electric vehicles. The Federal Government has set a goal to make half of all new vehicles sold in the U.S. in 2030 zero-emissions vehicles. It is projected that there will be 26.4 million electric vehicles on U.S. roads in 2030. One concern regarding the adoption of electric vehicles is the ability of power systems to accommodate their high-power demand. Another concern is the present high costs of electric vehicles, which make them unaffordable for most of the country’s population. This project contributes a solution to address both the concerns. First, it contributes to developing advanced intelligent demand response programs, which have been recognized as being effective in shaving peak demand of power systems (including the demand by electric vehicles), thereby reducing the system operation cost and cutting costs by deferring equipment upgrade and investment. Such intelligent demand response programs can potentially save billions of dollars annually. Second, the project develops intelligent algorithms that enable transactions between electric vehicles and power grids, where the vehicle owners can make considerable additional income by charging during off-peak hours and selling (i.e., discharging) power back to the power system during peak hours. The owners can earn thousands of dollars per year, thereby offsetting the high costs of electric vehicles and making them more affordable. Furthermore, the project supports underrepresented minorities and female students participating in high-level and high-quality research. Its overall outcomes increase sustainable development and economic competitiveness of the United States.The emphasis of this project is to advance artificial intelligence and machine learning algorithms for optimal management of electric vehicles interactions with the electric power grid. First, a hierarchical forecasting framework that is scalable and distributable is developed using cellular computational networks. Electric vehicle charging (Grid-to-Vehicle) and discharging (Vehicle-to-Grid) potential transactions are forecasted. Secondly, a hierarchical architecture-based methodology for scalable demand response with electric vehicles is developed. The hierarchical demand response architecture overlaying the physical hierarchy of the power system allows for decomposing the demand response to tackle the electric vehicle’s problem and solve it in a distributed manner. The computational time required to solve this optimization problem using this framework is only dependent on the number of levels in the hierarchical architecture. Thirdly, an adaptive critic design approach based on combined concepts of approximate dynamic programming and reinforcement learning is created for utilizing the capabilities of the electric vehicle battery systems for optimal reactive power compensation and voltage control on the distribution system. This is essential to maintain grid security and reliability as the number of electric vehicles penetrating the electric power distribution system rapidly grows to millions over the next few decades.This project is jointly funded by the CISE MSI Research Expansion and the Established Program to Stimulate Competitive Research (EPSCoR).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.

紧迫的挑战，例如气候变化和减少碳排放所需的必要挑战，需要从汽油动力的车辆过渡到电动汽车。联邦政府设定了一个目标，是在2030年在美国出售的所有新车辆中的一半。预计2030年美国道路上将有2640万辆电动汽车。采用电动汽车的一个担忧是电力系统适应其高功率需求的能力。另一个问题是目前的电动汽车成本高昂，这使得它们在该国大多数人口中无法承受。该项目为解决这两个问题的解决方案做出了贡献。首先，它有助于制定先进的智能需求响应计划，这些计划被认为有效地刮去了电力系统的峰值需求（包括电动汽车的需求），从而降低了系统运行成本和通过推迟设备升级和投资来降低成本。这种智能需求响应计划可能每年节省数十亿美元。其次，该项目开发了智能算法，这些算法可以使电动汽车和电网之间的交易，在此期间，车主可以在非高峰时段充电并在高峰时段销售（即，将电力释放回电力系统）来赚取大量收入。业主每年可以赚取数千美元，从而抵消电动汽车的高昂成本并使其更实惠。此外，该项目支持代表性不足的少数群体和女学生参加高级和高质量研究。它的整体成果提高了美国的可持续发展和经济竞争力。该项目的重点是推进人工智能和机器学习算法，以最佳地管理电动汽车与电力电网的交互。首先，使用蜂窝计算网络开发可扩展和分配的分层预测框架。预计电动汽车充电（车辆到车辆）和排放（车辆到网格）的潜在交易已预测。其次，开发了基于层次结构的基于层次结构的方法，用于对电动汽车的可扩展需求响应。覆盖电力系统物理层次结构的层次需求响应架构允许分解需求响应以解决电动汽车的问题并以分布式方式解决。使用此框架解决此优化问题所需的计算时间仅取决于分层体系结构中的级别数量。第三，创建了一种基于近似动态编程和强化学习概念的自适应关键设计方法，用于利用电动汽车电池系统的功能，以实现最佳的反应性电源补偿和分配系统的电压控制。这对于保持网格安全性和可靠性至关重要，因为在接下来的几十年中，穿透电力发电系统的电动汽车数量迅速增长到数百万。该项目由CISE MSI研究的扩展和既定的竞争性研究（EPSCOR）的既定计划共同资助（EPSCOR）。这一奖项反映了NSF的法定任务和经过评估的范围。