CPS Medium: Autonomous Control of Self-Powered Critical Infrastructures

该网络物理系统（CPS）项目将开发新型传感、驱动和嵌入式计算技术，使民用基础设施在面对动态负载时能够做出响应、有弹性和自适应。这种技术需要输送电力，通常是通过外部电网，或者通过使用电池存储。然而，在极端负载情况下，电网供电可能不可靠，电池必须定期充电或更换。在这个项目中开发的技术的新颖之处在于，它们通过储存和再利用外部负载注入基础设施的能量来为自己供电。该项目侧重于三种应用：(1)城市雨水网络，利用水文流动产生的电力主动控制水位以防止洪水；(2)建筑物在地震和大风中主动控制变形，利用振动产生的电力；(3)海洋淡化系统，利用波浪产生的电力主动控制抽水速率。该项目包含每个应用程序的实验活动。它还包含一个分析组件，专注于控制算法的开发，以最大限度地提高技术的性能。教育拓展活动包括针对本科生和研究生的课程模块和研究体验，以及针对高中生的研讨会。自供电基础设施的控制算法必须明确优化发电和性能目标之间的平衡。该项目将为自供电基础设施技术创新新的模型预测控制算法，使其在不耗尽能量的情况下实现最佳性能。这些算法将在实验中得到验证，适用于所有三种应用。目前还没有一个理论可以用于可扩展到大型复杂系统的自供电系统的最优控制。在这里进行的研究将增强模型预测控制理论的最新进展，从而形成该领域的新知识体系。挑战包括：(1)创新优化算法，以应对最优自驱动控制问题的固有非凸性；(2)开发有效的技术来处理目标应用的动态随机性；(3)综合计算可处理的控制器，但也能最优地补偿动力系统中复杂的传输损耗和约束；(4)系统技术的推导，以确保控制器对系统模型中的不确定性和干扰的鲁棒性。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 Cyber Physical Systems (CPS) project will develop novel sensing, actuation, and embedded computing technologies that allow civil infrastructures to be responsive, resilient and adaptive in the face of dynamic loads. Such technologies require delivery of electrical power, typically either via an external power grid, or through the use of battery storage. However, grid power may be unreliable during extreme loading events, and batteries must be periodically recharged or replaced. The novelty of the technologies developed in this project is that they power themselves, by storing and reusing energy injected into the infrastructure by external loads. The project focuses on three applications: (1) urban stormwater networks that actively control water levels to prevent flooding, using power generated from the hydrologic flows, (2) buildings that actively control their deformations during earthquakes and high winds, using power generated from vibrations, and (3) ocean desalination systems that actively control pumping rates, using power generated from waves. The project contains an experimental campaign for each application. It also contains an analytical component, focused on the development of control algorithms to maximize the performance of the technologies. Educational outreach activities include class modules and research experiences for undergraduate and graduate students, as well as a workshop for high school students. Control algorithms for self-powered infrastructures must explicitly optimize the balance between power generation and performance objectives. This project will innovate new Model Predictive Control algorithms for self-powered infrastructure technologies, such that they achieve the best performance possible while not running out of energy. These algorithms will be validated experimentally, for all three applications. There is presently no existing theory for optimal control of self-powered systems that is scalable to large and complex systems such as the ones under consideration. The research to be conducted here will augment recent advances in Model Predictive Control theory, to result in a new body of knowledge in this area. Challenges include: (1) innovation of optimization algorithms that can contend with the inherent nonconvexity of optimal self-powered control problems; (2) development of effective techniques for handling the stochastic nature of the dynamics for the target applications; (3) synthesis of controllers that are computationally tractable, but which also optimally compensate for the complex transmission losses and constraints in the power trains; (4) the derivation of systematic techniques for ensuring the robustness of the controllers, to uncertainties in the system model and disturbances.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.