RESUMO
Quantifying the uncertainty of stormwater inflow is critical for improving the resilience of urban drainage systems (UDSs). However, the high computational complexity and time consumption obstruct the implementation of uncertainty-addressing methods for real-time control of UDSs. To address this issue, this study developed a machine learning-based surrogate model (MLSM) that maintains high-fidelity descriptions of drainage dynamics and meanwhile diminishes the computational complexity. With stormwater inflow and controls as inputs and system overflow as the output, MLSM is able to fast evaluate system performance, and therefore stochastic optimization becomes feasible. Thus, a real-time control strategy was built by combining MLSM with the stochastic model predictive control. This strategy used stochastic stormwater inflow scenarios as input and aimed to minimize the expected overflow under all scenarios. An ensemble of stormwater inflow scenarios was generated by assuming the forecast errors follow normal distributions. To downsize the ensemble, representative scenarios with their probabilities were selected using the simultaneous backward reduction method. The proposed control strategy was applied to a combined UDS of China. Results are as follows. (1) MLSM fit well with the original high-fidelity urban drainage model, while the computational time was reduced by 99.1%. (2) The proposed strategy consistently outperformed the classical deterministic model predictive control in both magnitude and duration dimensions of system resilience, when the consumed time compatible is with the real-time operation. It is indicated that the proposed control strategy could be used to inform the real-time operation of complex UDSs and thus enhance system resilience to uncertainty.
RESUMO
The active control of urban drainage systems (UDSs) is playing an increasingly important role in the world threatened by urban flooding and associated disasters caused by insufficient drainage capacity. However, little research has recognized the importance of the optimal use of in-pipe storage space. To address this issue, the use of the in-pipe storage capacity was optimized in this study. A novel approach, that is, dynamic programming with successive approximation considering the time lag of flow routing (DPSA-TL), was developed to determine the control policies, in addition to the commonly used passive, rule-based control (RBC), and evolutionary algorithm (EA) strategies. A real-life urban catchment considering flooding control and combined sewer overflow (CSO) reduction was used as the case study. First of all, the potential benefit of maximizing the use of in-pipe storage space was tested using the four control strategies in three storm events, including a 3-year, 2-hour design (46.5 mm), a 5-year, 2-hour design (56.0 mm) and a 7-h historical (152.5 mm) storm events. Results indicate that DPSA-TL performed best in all cases. Without compromising the goal of flooding control, it provided 16.5%, 12.6%, and 3.0% reductions in CSO volume for the three storm events when compared with the passive strategy. Due to the limited capacity of in-pipe storage, the relative improvement diminished as the total rainfall depth increased. Then, control strategies were further applicated to the real-time operation. DPSA-TL was found to be the best alternative for CSO control, with the CSO volume reduced by 14.7%, 11.4%, and 2.5% in the three storm events, respectively. The findings suggest that the performance of UDS can be significantly improved by optimizing the use of in-pipe storage capacity, and the proposed method is effective in the offline optimization and real-time control of UDSs.