基于可用水量预测的梯级水电站水库调度函数

Cascaded Reservoirs Operation Rules Based on Predicted Available Water

摘要: 以可用水量为状态变量的梯级水电站水库调度函数由于状态变量存在估算误差，在结合预报来水应用时，易造成应用效果欠佳。为此，提出以可用水量预测值为状态变量的调度函数及其优化模型。该调度函数以面临时段有效库容叠加季节性自回归滑动平均求和模型(ARIMA)预报来水量作为可用水量预测值，以此为自变量构建梯级水电站水库调度函数。在多元回归参数估计的基础上，建立以模拟多年平均发电量最大为目标，以保证率及保证出力破坏程度为约束条件的调度函数参数优化模型。澜沧江中下游梯级水库群的实例研究表明，新的调度函数能有效考虑实际操作中的径流预测不确定性问题，得到更优的模拟调度结果。

Abstract: The traditional operation function for cascaded reservoirs takes available water as the state variable. However, poor simulation results may be caused by estimation errors when using predicted inflow. For this reason, with predicted available water considered as the state variable, an operation function and an optimization model are proposed in this paper. In this operation function, the predicted value equals existing storage in the current stage plus local inflow forecasted by seasonal ARIMA model, and the parameters are obtained by multiple regression method. The proposed optimization model aims to maximize annual average generating capacity with consideration of system reliability and extreme failure. A case study of Lancang River cascaded reservoirs shows that the proposed operation function can effectively consider the uncertainty of forecasted run-off and obtain a better solution than the traditional one.