CAREER: Toward a theory for smart stormwater systems

美国的许多城市都有大型的雨水盆地、运河和地下管道网络。当下雨时，这些基础设施将水和污染物从建筑物转移到附近的河流和溪流。在许多城市，这些雨水系统要么太小，要么太旧，导致了危险的洪水泛滥，并污染了下游的河流和湖泊。升级基础设施是一个代价高昂的选择。该项目将展示如何更有效地利用现有的雨水系统，并通过使用无线传感器、阀门和泵来“更智能”地解决这一全国性问题，而不是依靠新建筑。这种方法将使雨水系统自动控制，以减少洪水和改善水质。为此，PI将研究将“智能”嵌入雨水系统的计算方法，以帮助降低改善美国基础设施的成本，同时保护公众健康和环境。该项目还将为新一代技术人员制定培训计划，这些技术人员将部署和维护智能雨水系统。PI将提高对通过城市流域实时控制可以实现的水流和水质效益的基本认识。这项研究的结果将导致一个理论和概念框架，开始研究自主技术对整个流域的影响。该提案的研究假设是，“智能”雨水系统将大大缩小管理径流污染所需的基础设施的规模。这一假设将通过三个目标进行检验：(1)研究基于动力系统理论和强化学习的新算法，以控制城市尺度雨水系统的流量；(2)研究整个流域尺度上的分布式实时水质控制，重点是去除沉积物和溶解污染物；(3)探索如何统一相互竞争的水流和水质目标，以实现城市流域的实时弹性管理。这项研究的结果将使整个流域得到控制，类似于大型分布式处理厂，同时极大地提高了对“智能”水系统的好处、权衡和性能限制的基本理解。所有的发现、数据和算法将通过一个开源门户网站与其他研究人员和实践者免费共享。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。

Many cities across the United States maintain large networks of stormwater basins, canals, and underground pipes. When it rains, this infrastructure moves water and pollutants away from buildings to nearby rivers and streams. In many cities, these stormwater systems are too small or too old, resulting in dangerous flooding and pollution of downstream rivers and lakes. Upgrading infrastructure is a costly alternative. Instead of relying on new construction, the PI will demonstrate how existing stormwater systems can be used more efficiently and made "smarter" through the use of wireless sensors, valves, and pumps to address this national issue. This approach will enable automatic control of stormwater systems to reduce flooding and improve water quality. To this end, the PI will investigate computational methods to embed stormwater systems with "intelligence," helping to reduce the cost of improving US infrastructure while protecting public health and the environment. The PI will also develop training programs for a new generation of technicians, who will deploy and maintain smart stormwater systems. The PI will improve the fundamental understanding of the water flow and water quality benefits that can be achieved through the real-time control of urban watersheds. Results of this research will lead to a theoretical and conceptual framework to begin studying the impacts of autonomous technologies on the scale of entire watersheds. The research hypothesis of this proposal is that "smart" stormwater systems will vastly shrink the size of infrastructure required to manage runoff pollution. The hypothesis will be tested across three objectives: (1) investigate new algorithms, based on dynamical systems theory and reinforcement learning to control flows across city-scale stormwater systems, (2) study the distributed real-time control of water quality across the scale of entire watersheds with a focus on removing of sediment-bound and dissolved pollutants, and (3) explore how competing water flow and water quality goals can be unified to resiliently manage urban watersheds in real-time. Results of this research will allow entire watersheds to be controlled, akin to large distributed treatment plants, while vastly improving the fundamental understanding of the benefits, tradeoffs, and performance limits of "smart" water systems. All findings, data and algorithms will be shared freely with other researchers and practitioners through an open source web portal.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.