澜沧江中下游梯级水库群多目标调控模型与方法

Focusing on the key theory and technical challenges in the multi-objective optimal operation model and method of cascade reservoirs under the condition of runoff variation, this presented project, which is aimed at the Lancang River in Southwest of China, firstly analyzes the hydrologic regime and its evolution trend under changing environment in the Lancang River basin. Then a calculation method of ecological water demand, driven by the hydrological-ecological response relationship, will be investigated. Furthermore, the mathematical abstraction and description is tried to be captured within the analysis of water resources control indicators of the key river section in the basin. Afterwards, the water supply-power generation-ecological environment multi-objective optimal model will be established, which combines the water supply, water demand of ecological environment and power generation of cascade reservoirs. And then the method of multi-objective risk analysis and multi-attribute risk decision-making will be addressed, which largely helps to evaluate the regulation risks scientifically and select schemes judiciously. Finally, adaptive operation rules will be extracted by using the stochastic scheduling simulation and data mining technology. As it goes on, it will ultimately provide a theoretical basis and technological support for the water supply-power generation-ecological environment multi-objective optimal operation in the Lancang River basin.

围绕径流变化条件下流域梯级水库群多目标调控面临的关键理论与技术难题，以澜沧江中下游梯级水库群为重点研究对象，分析变化环境下流域水文情势演变趋势，研究基于水文-生态响应关系的河流生态环境需水过程推求方法，并根据流域内各区域用水需求计算河流关键断面水资源控制指标，实现对调控域边界的数学化描述；在流域供水和生态环境用水需求分析的基础上，结合梯级电站发电调度需求，通过调度目标量化，建立流域供水-发电-环境多目标均衡的梯级水库群联合优化调度模型；研究基于改进帕累托支配关系的多目标求解方法以实现模型的高效求解；研究流域梯级水库群多目标调度风险分析方法与多属性风险决策方法，实现对梯级水库群调控风险的科学评估和对调控方案的优选评价；最后，运用随机调度模拟和数据挖掘方法，提取梯级水库群适应性调度规则，为径流变化条件下澜沧江中下游梯级水库群供水-发电-环境多目标优化调控提供理论依据和技术支撑。

项目以澜沧江中下游梯级水库群为研究对象，围绕库群多目标适应性调控面临的关键理论与技术难题开展了系统深入研究，取得如下研究进展和成果：① 针对传统水库调度对生态需求量化考虑不足的问题，基于RVA框架推求了河流断面适宜生态流量上下限，以生态溢缺水量最小为目标，建立了梯级水库群发电-环境多目标优化调度模型；同时，针对梯级调度系统的复杂约束边界，研究提出了可行域边界的多重复杂约束自适应集成方法，推求并缩小了梯级水库群时段出力、运行水位和下泄流量的优化调控范围，从而提高了算法寻优效率。② 针对流域梯级水库群发电-环境多目标优化调度模型的求解问题，研究了基于支配-分解的区域搜索机制，提出了基于区域搜索的多目标算法，实现调度模型的高效求解，获得了关于各调度目标的非劣调度方案集，量化分析了不同调度目标间的竞争与冲突关系。③ 针对确定性优化模型和方法难以应用于梯级水库群中长期调度决策的问题，研究提出了一种考虑多重不确定性的水库调度规则提取方法；采用基于贝叶斯神经网络的深度学习方法，挖掘了决策变量与预报概率来水、水位等关键状态变量之间的映射关系，提取了同时考虑来水不确定性以及模型参数不确定性的澜沧江流域梯级水库群调度规则。④ 收集整理了澜沧江流域的地形、土地利用、土壤、气象、水文水质等基础数据，建立了基于SWAT模型和CE-QUAL-W2模型的澜沧江流域生源物质迁移过程模拟模型；针对天然情况、水库常规调度规则、发电最优调度规则和生态最优调度规则四种典型调控方案开展了生源物质迁移过程模拟，初步分析了总氮、总磷在澜沧江中下游河段迁移规律。项目成果可为径流变化条件下澜沧江中下游梯级水库群多目标适应性调控提供理论依据和技术支撑。