基于多目标遗传算法的巨型水库群发电优化调度  

Optimal operations of multi-reservoir system using multi-objective genetic algorithms

作　　者：魏加华[1] 张远东[2] 

机构地区：[1]清华大学水沙科学与水利水电工程国家重点实验室,北京100084 [2]水利部水资源管理中心,北京100053

出　　处：《地学前缘》2010年第6期255-262,共8页Earth Science Frontiers

基　　金：国家"十一五"科技支撑计划项目(2009BAC56B03)

摘　　要：低碳时代,水电作为清洁可再生能源,节能发电调度为梯级水电联合调度运行带来了难得的历史机遇,如何有效地开展梯级水库群优化调度,充分合理利用水能资源,是流域梯级电站管理迫切需要解决的问题。文章在对清江梯级与三峡梯级径流特征分析的基础上,建立了以水库群整体发电量最大和弃能最小为目标的梯级调度模型。根据电力系统对三峡、葛洲坝、水布垭、隔河岩、高坝洲水电站的要求,用1951—2002年的月径流资料和典型年的日径流资料进行长期和短期优化调度,采用多目标遗传算法得到清江3个电站、三峡-葛洲坝2电站单独运行和联合优化调度发电指标,5库联合优化调度系统多年平均发电量增加约21亿kW.h;短期(日)优化调度较长期(月时间尺度)优化调度发电效益有进一步提升,增发电量约9.68亿kW.h。研究表明,充分利用清江和三峡梯级实际运行位置相近、水力联系紧密、互补性强的特点,统一安排电站机组运行模式,合理分配机组出力,可在来水量相同的条件下获得更大的效益。In low-carbon era,hydropower as clean and renewable energy,energy-conservation based generation dispatching is an important measure to reduce energy consumption in China and will bring a rare historical opportunity for the hydropower joint dispatching operation.How to effectively carry out cascade optimal operation of reservoirs,and to fully make rational use of water resources are urgent needs of cascade hydropower station management.Based on the analysis of inflow characteristics of the Qingjiang and Three Gorges cascades,a multi-objective optimal model for hydropower dispatching is established.The two objective functions of optimization are to maximize hydropower generation and to minimize of the total discarded water for both independent and joint dispatching scenarios.The multi-objective optimal model for hydropower generation system is a large-scale dynamic nonlinear constrained optimization problem.Multi-objective genetic algorithm（MOGA） is employed to solve the model.Data of monthly average runoff from 1951 to 2002 and daily runoff of typical year are used to long-term and short-term optimization operation,respectively.The optimization results show that joint dispatching can increase power generation：（1） for long-term operation,the hydropower generation of the two cascades will averagely increase about 2100 GW·h without considering peak regulation;（2） for short-term operation based on long-term dispatching,the power generation of the two cascades will reincrease about 968 GW·h.To achieve more power benefit,it is vital to take full advantage of complementarities in geography,water resource and flood control.

关 键 词：水库调度 优化模型 多目标遗传算法 三峡梯级 清江梯级 

分 类 号：TV697.11[水利工程—水利水电工程]