采用"优化-拟合-再优化-随机仿真"技术框架研究梯级水库群联合优化调度函数:首先建立梯级水库群优化调度模型,通过改进的动态规划方法得到最优样本;然后对最优样本进行回归分析得到联合调度函数;并基于模拟调度结果采用非线性规划法直接修正调度函数;最后通过水文随机模拟技术生成径流系列,进一步检验、评价所得调度函数的效率。以长江上游大渡河梯级水库长期优化调度为对象进行实例研究,计算结果表明:对于1959-2000年实测径流系列,梯级水库群联合优化调度函数较常规调度年均发电量增幅达1.66%;对于Copula随机模型生成的水文径流系列,年均发电量提高了1.42%,经济效益显著。
Optimal operation rules for cascade reservoirs are derived in the framework of optimization, fitting, re-optimization and simulation. First, optimal data samples are generated using deterministic dynamic programming method to solve an optimization model, and they are used to establish operation rules with a regression analysis on them. Then, these rules are refined through simulations of the optimal operation and application of nonlinear programming, and verified with a series of synthetic inflow generated by Copula function. A case study of the Dadu river cascade hydropower plants shows that the proposed method is effective in improving the hydropower generation of cascade reservoirs.