An enhanced PSO-based approach for optimizing the flood control operation of reservoir networks

Abstract

[Objective] The flood control optimization scheduling of reservoir clusters plays a crucial role in flood management during heavy rain and flooding events. However, existing studies on improving the PSO algorithm often lack constraints and adjustments on the distance between particles and the optimal solution during the iteration process. Additionally, they fail to comprehensively consider both the downstream flood control targets and the safety of the reservoirs themselves during the optimization scheduling. [Methods] To better address the flood control optimization scheduling problem of reservoir clusters, an optimization model is established with the objective of maximizing peak shaving and minimizing the highest water level. The model focuses on four reservoirs in the Fei River Basin of Feixian County, Shandong: Longwangkou, Shangye, Xujiaya, and Shilan. The inertia weight and learning factors of the PSO algorithm are dynamically adjusted during the optimization process using trigonometric functions and Beta distributions. Additionally, the Central Limit Theorem is introduced to impose real-time constraints and regulation on the iterative process, further improving the PSO algorithm. The input conditions are the inflow of design floods with a recurrence interval of 100 years and 1000 years, and the optimization scheduling model is evaluated by considering flood control constraints and flood evolution. [Results] The result demonstrate that the larger the reservoir capacity, the more significant the peak shaving effect. Under the input conditions of a 100-year flood, the maximum discharge flow from the Xujiaya Reservoir was reduced by 559.62 m³/s compared to conventional scheduling, and by 279.81 m³/s compared to the standard PSO-optimized scheduling, achieving a peak shaving rate of 10.4%. The reservoir capacity was reduced by 6.4% compared to conventional scheduling and by 5.3% compared to the standard PSO optimization. Under the input conditions of a 1000-year flood, the maximum discharge flow from the Xujiaya Reservoir was reduced by 701.79 m³/s compared to conventional scheduling, and by 350.90 m³/s compared to PSO-optimized scheduling, achieving a peak shaving rate of 12.1%. The reservoir capacity was reduced by 9.2% compared to conventional scheduling and by 4.8% compared to PSO-optimized scheduling. [Conclusion] The result indicate that the proposed optimization scheduling model shows significant effects in achieving maximum peak shaving and minimizing water levels. The algorithm ensures effective precision and stability during the optimization process, demonstrating strong optimization performance and substantial practical application value.

Keywords

reservoir group / peak shaving criterion / improved PSO algorithm / optimization scheduling / influencing factors