基于聚合–分解的并联水库群防洪优化调度研究  

Optimization of Parallel Reservoirs for Flood Control Based on the Aggregation-Decomposition Method

作　　者：陈西臻 刘攀[1,2] 何素明 张旺[1,2] 李泽君[1,2] 张靖文[1,2] 

机构地区：[1]武汉大学水资源与水电工程科学国家重点实验室,湖北武汉 [2]水资源安全保障湖北省协同创新中心,湖北武汉 [3]广西壮族自治区水利电力勘测设计研究院,广西南宁

出　　处：《水资源研究》2015年第1期23-31,共9页Journal of Water Resources Research

基　　金：优秀青年科学基金(51422907);教育部新世纪优秀人才支持计划(NCET-11-0401);水利部公益性行业科研专项经费项目(201201051);中央分成水资源费项目(项目编号:1261430210028)

摘　　要：基于聚合-分解框架，构建了并联水库群防洪调度函数：在聚合步骤中，将总出库流量表示为聚合水库库容的分段线性函数；在分解步骤中，按各水库来水量比例分解总出库流量，得到各水库出库流量。从而构建了并联水库群防洪优化调度模拟模型，采用遗传算法进行模拟优化，识别分段线性函数的关键参数。以广西西江流域百色、龙滩、青狮潭三个并联水库为实例开展研究，计算结果表明，较常规调度，优化调度在率定期（共10场次洪水】和检验期【共5场次洪水】分别能使下游防洪控制点梧州站平均多削峰970m3／s和1320m3／s。因此，提出的方法对指导水库群防洪优化调度有一定价值。The optimal operating rules for parallel reservoirs were built based on the aggregation-decom- position method. The relationship between the total release of the reservoirs and the aggregation reservoir storage could be expressed by piecewise-linear operating rules in the aggregation step, and the release was decomposed to each parallel reservoir according to its ratio of inflow in the decomposing procedure. Then the optimization model of parallel reservoirs was built, and its key parameters were calibrated by the Genetic Algorithm. Three parallel reservoirs （the Baise, Long- tan and Qingshitan reservoirs） of the Xijiang River in China were selected for a case study. The op- timal flood operating rules were compared with the conventional operating rules, and the flood peak of reference station for flood control could be decreased by 970 m3/s and 1320 m3/s in cali- brated and validation periods, respectively. Thus, this method is beneficial for the reservoir＇s real- time operation to decrease the flood disasters.The optimal operating rules for parallel reservoirs were built based on the aggregation-decom- position method. The relationship between the total release of the reservoirs and the aggregation reservoir storage could be expressed by piecewise-linear operating rules in the aggregation step, and the release was decomposed to each parallel reservoir according to its ratio of inflow in the decomposing procedure. Then the optimization model of parallel reservoirs was built, and its key parameters were calibrated by the Genetic Algorithm. Three parallel reservoirs （the Baise, Long- tan and Qingshitan reservoirs） of the Xijiang River in China were selected for a case study. The op- timal flood operating rules were compared with the conventional operating rules, and the flood peak of reference station for flood control could be decreased by 970 m3/s and 1320 m3/s in cali- brated and validation periods, respectively. Thus, this method is beneficial for the reservoir＇s real- time operation to decrease th

关 键 词：分段线性 并联水库 聚合-分解 遗传算法 调度函数 

分 类 号：TV697.11[水利工程—水利水电工程]