水电站汛期防洪与发电协调控制理论与方法研究

With respect to the low utilization of hydropower resources, this project aims to improve the utilization efficiency of hydropower resources and mitigate the flood risk from forecast uncertainty, through the coordinated control of flood control and hydropower generation during flood season. First, the reservoir operation procedure is generalized to a two-stage rolling over system, and then the two-stage optimization model, to maximum the hydropower generation and minimize the flood control risk, is established based on the impact analysis of the uncertainty in hydrological forecast on flood risk. Combining the method of mathematical analysis and the marginal utility principle, economical characteristics of objective functions are illustrated, and also the optimality condition of the model is derived. Second, with the optimality condition, the scope of the forecast inflow that there exists competitive relationship between two objectives is derived quantitatively, and also the optimization decision within or outside the scope is derived. Following that the competitive mechanism and the conversion relationship between flood risk and hydropower generation during flood season is explored. And then, the influencing mechanism of forecast uncertainty and acceptable flood risk on optimal decision are explored. Lastly, the Nuozhadu Reservoir in Southwest and Nierji Reservoir in Northeast are taken as the case studies to validate the research results. The research in this program has certain theory and the practical application value, and provides a brand-new idea for dynamic control of flood limited water level.

针对水电站汛期弃水严重、水能利用率低的问题，本项目将展开水电站汛期防洪与发电双目标协调控制研究，利用预报信息动态控制共用库容提高水能资源利用率。首先，将水库汛期调度过程转化为当前与未来两阶段滚动决策过程，分析预报信息不确定性与防洪风险的关系，以防洪风险最小和发电量最大为目标，建立防洪与发电双目标两阶段协调控制优化对冲模型，结合边际效用理论，阐述目标函数经济学特性，提出双目标最优协调控制条件；然后，基于最优协调控制条件，揭示水电站汛期防洪与发电双目标的竞争关系与转化机制，提出两目标对冲的来水范围确定方法，定量给出范围内外的最优决策，制定优化对冲规则，并阐明预报不确定性和防洪标准对两目标竞争关系的影响机理。最后，选择南方糯扎渡电站和北方尼尔基电站验证本项目的研究成果。本项目的研究具有一定的理论和实际应用价值，为汛限水位动态控制提供了一条新的思路。

利用预报信息对水库共用库容实施动态控制是提高洪水资源利用率、水能利用率，增大防洪效益的有效手段。本项目针对水库汛期共用库容动态控制的关键问题，分析降雨预报与洪水预报的可利用性及其不确定性，挖掘预报不确定性对防洪风险的影响，分析水库利用预报信息进行预蓄预泄和预泄回蓄调度时各目标竞争关系，最终建立水库汛期共用库容动态控制理论。创新成果包括：系统分析了集合降雨预报不同集成方法和不同利用形式的预报精度，为集合降雨预报信息在水库调度的应用提供技术支撑。构建了基于机器学习的洪水预报模型，提出了基于方差分解理论的不确定性量化方法，解析了样本集划分、机器学习方法选择和模型架构设计对洪水预报不确定性的影响随预见期和流量量级变化的规律。提出了降雨及洪水预报信息单独应用、耦合应用时水库防洪风险的计算方法，基于贝叶斯理论挖掘了不同降雨预报量级下洪水预报信息对水库防洪风险影响的规律。分别建立了水库汛期采用预泄回充方法增大防洪效益、采用预蓄预泄方法增大发电效益的模型，解析了预蓄调度时防洪与发电之间的竞争关系，揭示了预泄调度时下游防洪风险、水库的防洪效益与水库回蓄风险之间的竞争与转化机制。共发表论文12篇，其中SCI/EI检索9篇，包括JWRPM2篇，WARM3篇。