市场环境下基于奔德斯分解的年末水位优化策略及模型

Study on strategy and modeling of biennial stochastic scheduling for full scenario cascade hydropower based on Benders decomposition

在市场环境下,水电站年末水位优化需要考虑来水、负荷等不确定性因素,提出了年末水位优化与短期市场出清一体化决策框架,在满足各种场景约束条件下,实现了收益最大化。基于该框架,构建以目标年的水电收益最大化为上层目标、以各种场景下市场出清福利最大化为下层目标的双层规划模型,其中上层决策优化确定水电厂年末预留水位,下层决策优化确定各场景、各时段下的市场出清价,运用随机规划的互补性理论转化为等式约束数学规划模型(MPEC),并运用强对偶理论,将模型中的非线性项进行线性化,形成大规模混合整数线性规划模型(MILP)。提出了主、子问题一体控制的最优奔德斯(Benders)分解方法,确保分解协调的最优性与高效性。最后,算例验证了模型和求解方法的有效性。

In power market environments, inflow uncertainty and load uncertainty should be considered in scheduling optimization of hydropower. This paper formulates a framework of integrated decision-making for end-year water level optimization and short-term market clearing to maximize profits and improve economy and scheduling security. Within this framework, we constructed a bi-level programming model with the upper and lower levels for maximizing the profits and social welfare through optimization of end-year water and the market clearing prices, respectively. This model was converted into a mathematical program with equilibrium constraints （MPEC） using the complementarity of stochastic programming, and it was linearized by the strong duality theorem into a large-scale mixed integral linear programming. We also present an optimal Benders decomposition for the master and subproblems to ensure optimality and high performance. Finally, a case study is used to demonstrate the proposed model and its solution procedure.