Real‐Time Control of Urban Headwater Catchments Through Linear Feedback: Performance, Analysis, and Site Selection
Real‐Time Control of Urban Headwater Catchments Through Linear Feedback: Performance, Analysis, and Site Selection
复制标题
通过线性反馈实时控制城市水源集水区：性能、分析和选址
DOI：
10.1029/2018wr022657
复制
发表时间：
2018
影响因子：
5.4
通讯作者：
B. Kerkez
中科院分区：
文献类型：
--
作者：
B. Wong;B. Kerkez
城市流域的实时控制现在正通过新一代“智能”和互联技术实现。通过用传感器和阀门改造雨水系统，使整个流域动态适应个别风暴成为可能。提出了一种流域尺度控制算法，该算法将城市流域抽象为一个线性积分器时滞动态系统，利用流域物理特征对其进行参数化，然后利用线性二次型调节器对网络流量进行控制。该方法在密歇根州安阿伯市一个4平方公里的城市水源集水区进行了模拟，展示了雨水系统的收益，该系统可以自适应地平衡洪水缓解和流量减少。我们引入了等效分析，并举例说明了受控流域在大事件（30年风暴）中的表现，以表明非受控被动流域只能在较小事件（10年风暴）中与之匹配。对于这些较小的事件，控制存储节点（池塘、流域和湿地）的存储量最多可以减少50%，但仍能达到控制流域的相同性能。还进行了控制器放置分析，从而在广泛的设计风暴范围内模拟受控站点的所有可能组合。我们表明，在流域中可能不需要控制每个存储节点，而是在我们的案例研究中，可以协调控制整个流域的一小部分（30%），以实现与完全控制系统相匹配的结果，即使在长期降雨记录和噪声传感器测量下进行测试时也是如此。
The real‐time control of urban watersheds is now being enabled by a new generation of “smart” and connected technologies. By retrofitting stormwater systems with sensors and valves, it becomes possible to adapt entire watersheds dynamically to individual storms. A catchment‐scale control algorithm is introduced, which abstracts an urban watershed as a linear integrator delay dynamical system, parameterizes it using physical watershed characteristics, and then controls network flows using a Linear Quadratic Regulator. The approach is simulated on a 4‐km2 urban headwater catchment in Ann Arbor, Michigan, demonstrating the gains of a stormwater system that can adaptively balance between flood mitigation and flow reduction. We introduce an equivalence analysis and illustrate the performance of the controlled watershed across large events (30‐year storms) to show the uncontrolled passive watershed can only match it during smaller events (10‐year storm). For these smaller events, the storage volume of the controlled storage nodes (ponds, basins, and wetlands) could be reduced as much as 50% and still achieve the same performance of the controlled watershed. A controller placement analysis is also carried out, whereby all possible combinations of controlled sites are simulated across a wide spectrum of design storms. We show that the control of every storage node may not be needed in a watershed, but rather that in our case study a small subset (30%) of the overall watershed can be controlled in coordination to achieve outcomes that match a fully controlled system, even when tested across a long‐term rainfall record and under noisy sensor measurements.