PFI:BIC Self-Learning Algorithms for Advancement of Smart Stormwater Green Infrastructure Systems

在全国范围内，一个紧迫的社会和环境问题是雨水对水体损害的影响，特别是在城市和郊区流域。例如，雨水在流过陆地进入溪流时，会释放出金属和营养物质等污染物，从而降低水质。雨水在联合下水道系统中造成溪流侵蚀、沉积、洪水和溢流。为了解决这些问题，市政当局必须采用创新技术，如生物滞留、人工湿地和植被屋顶等绿色基础设施（GI）系统。尽管地理标志系统具有优势，但由于技术和人为因素，其采用速度缓慢。这些系统还不是动态的，不能适应季节变化，并且通常只能实现一个性能目标，从而导致高实施和维护成本。其他潜在因素包括运营和维护问题、政策和融资问题、缺乏不同利益相关者的支持以及投资回报不明确。这项研究将开发“智能”（即高效、主动和自我学习）雨水服务系统。“智能”系统使用传感器和人工生成的数据来简化GI维护程序，从而在性能、预测和故障预防方面降低成本，提高效率。研究活动的更广泛影响是在各个处理尺度上改善雨水管理。研究活动将导致更有效和更经济的地理标志，通过在受损水体中减轻洪水和改善水质方面取得重大进展，证明其必要性和对社会和所有流域利益攸关方的益处。底层技术为直接将基础设施连接到涉众提供了独特的机会，系统数据在基于云的数据管理系统中传输、存储和处理，并作为web服务发布。该项目采用了一种基于研究的方法，将应用与社会技术系统相结合。这一结果将通过最佳地利用GI系统中的所有物理过程（即滞留、渗透、蒸散）来实现，该系统使用传感器和控制系统，与实时天气和系统条件、预测数据和社交媒体相结合。Villanova大学的GI系统（绿色屋顶，建造的雨水湿地，生物保持雨水花园）将配备传感器（例如，土壤湿度，水位，温度和溶解氧），以及自动控制结构（例如，阀门或闸门），提供动态控制算法，在降雨期间和之后进行最佳操作。该GI系统将通过物理计算动态链接到具有实时可视化、数据可访问性、质量保证和实时控制的平台技术。整个自动化系统将在三个时间尺度上运行：1)降雨期间的小时，2)降雨后的几天，以及3)季节性尺度。所有的控制算法都将用于最大限度地储存雨水和改善水质。这些目标会因季节和气候带而异。主要合作伙伴包括维拉诺瓦大学（Villanova， PA）的土木与环境工程系和计算机科学系；宾夕法尼亚大学设计学院（Philadelphia， PA）；以及行业合作伙伴Geosyntec Consultants（马萨诸塞州波士顿）。更广泛的合作伙伴包括法国巴黎的威立雅；德克萨斯州奥斯汀市；区环境署；内布拉斯加州奥马哈市；费城水务局。

A pressing social and environmental issue on a national scale is the effect of stormwater on waterbody impairment, particularly in urban and suburban watersheds. For example, stormwater can diminish water quality by discharging pollutants like metals and nutrients as it runs over land and into the streams. Stormwater causes stream erosion, sedimentation, flooding and overflows in combined sewer systems. To combat these issues, municipalities must adopt innovative technologies, such as Green Infrastructure (GI) systems like bioretention, constructed wetlands, and vegetated roofs. Despite the advantages of GI systems, its adoption has been slow due to technological and human factors. These systems are not yet dynamic, cannot adapt to seasonal changes and are often able to accomplish only one performance goal resulting in high implementation and maintenance costs. Other potential factors include operation and maintenance issues, policy and financing issues, lack of buy-in from different stakeholders, and unclear return on investment. This research will develop "smart" (i.e., efficient, active and self-learning) stormwater service systems. "Smart" systems use sensor- and human-generated data to streamline GI maintenance programs to be less costly and more effective in performance, prediction and failure prevention. The broader impacts of the research activities are the improvement of stormwater management across treatment scales. The research activities will lead to more efficient and economical GI that demonstrates its need and benefit to society and all watershed stakeholders by making major progress towards flood mitigation and water quality improvement in impaired waterbodies. The underlying technologies allow for unique opportunities to directly connect infrastructure to stakeholders, system data is transmitted, stored, and processed in cloud-based data management systems and published as web services. This project enlists a research-based approach that integrates application with the socio-technical system. This outcome will be achieved by optimally using all physical processes in a GI system (i.e., detention, infiltration, evapotranspiration), which uses sensors and controls integrated with real-time weather and system conditions, forecast data, and social media. Villanova University GI systems (green roof, constructed stormwater wetland, bioretention rain garden) will be equipped with sensors (e.g., soil moisture, water level, temperature and dissolved oxygen), as well as automated control structures (e.g., valves or gates), providing dynamic control algorithms that optimally operate during and after rain events. This GI system will be dynamically linked to a platform technology with real time visualization, data accessibility, quality assurance and real time control through physical computing. The entire automated system will operate at three time scales: 1) hours during, 2) days after the rain event, and 3) seasonal scale. All control algorithms will be geared to maximize storage available for stormwater and water quality improvement. These goals will vary across season and climate zones. The primary partners include Department of Civil and Environmental Engineering, and Department of Computing Sciences from Villanova University (Villanova, PA); University of Pennsylvania School of Design (Philadelphia, PA); and Geosyntec Consultants, an industry partner (Boston, MA). The broader context partners include Veolia, Paris France; City of Austin, Texas; District Department of the Environment; City of Omaha, Nebraska; and Philadelphia Water Department.