An improved dynamic programming algorithm by integrating discrete differential and successive approximation for real...

Area of Science:

• Environmental Engineering
• Water Resource Management
• Computational Fluid Dynamics

Background:

• Urban drainage systems (UDSs) require enhanced storage and discharge for flood control and pollution prevention.
• Real-time orifice operation is critical for UDS performance, yet current methods lack precision and high dimensionality.
• Optimizing UDS storage-drainage capacity necessitates improved orifice control precision and reduced optimization dimensionality.

Purpose of the Study:

• To develop a novel optimization algorithm for enhancing UDS storage-drainage capacity.
• To improve orifice control precision and reduce optimization dimensionality in UDS.
• To demonstrate the effectiveness of the new algorithm in reducing combined sewer overflow (CSO) and improving UDS resilience.

Main Methods:

• Coupling Discrete Differential Dynamic Programming (DDDP) with Successive Approximation (SA) to create the DDDP-SA algorithm.
• Utilizing successive approximation to reduce problem dimensionality and DDDP for optimal solution search.
• Comparing DDDP-SA with passive, rule-based control (RBC), and DPSA-TL strategies using data from Yueyang, China.

Main Results:

• DDDP-SA reduced combined sewer overflow (CSO) volume by 1.84% to 11.03% compared to passive strategy across rainfall events.
• CSO mitigation effectiveness decreased with increasing rainfall intensity but remained superior in real-time control (RTC) scenarios.
• DDDP-SA improved UDS control precision and adaptability, increasing orifice usage duration and opening standard deviation by 4% to 8% compared to DPSA-TL.

Conclusions:

• The DDDP-SA algorithm offers an efficient approach for real-time optimization and fine-grained orifice operation in UDS.
• This methodology strategically utilizes the pipe network's storage capacity, significantly enhancing UDS resilience.
• The study highlights the importance of advanced algorithms for improving urban flood control and pollution prevention in UDS.