基于规律的多年调节水库年末消落水位预测模型  

Forecast model of year-end level for multi-year storage regulation reservoirs based on operation trends

作　　者：石萍[1] 纪昌明[2] 李继伟[2] 刘蕊鑫 

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,武汉430072 [2]华北电力大学可再生能源学院,北京102206

出　　处：《水力发电学报》2014年第2期58-64,共7页Journal of Hydroelectric Engineering

基　　金：国家自然科学基金资助项目(51279062;51179130);高校基本业务费重点项目(12zx04)

摘　　要：本文首先根据金沙江中游梯级水电站优化调度结果,分析了其优化调度规律。分析结果表明,金沙江中游梯级水电站的优化调度规律随来水的丰枯呈规律性衍变,各年的规律都由"保证出力"、"递增流量"、"不弃流量"三个基本元素组合而成。然后基于优化调度规律,分析了龙盘水库年末消落水位的影响因子,建立了龙盘水库年末消落水位预测模型。结果表明,本文建立的模型,机理清晰,精度较高,水库模拟模型和线性规划模型外延性优于人工神经网络模型。本文研究成果丰富了水库调度的机理研究,对金沙江中游梯级水电站年调度计划的制订具有指导意义,对其它"1库N梯"型梯级水电站调度规律的理论与应用研究具有参考价值。This paper presents optimal operation results and analyzes optimal operation rules for the cascade hydropower stations located in the Jinsha River middle reaches. Results show that in high and low flow seasons the optimal operation trends of these cascade stations vary with the inflow changes and that these trends in each year could be described by using a unique combination of three basic elements, i.e. guarantee capability, increasing flow, and no-surplus flow. By using operation trends, a year-end level forecast model was developed through analysis on the impact factors of year-end level in Longpan reservoir. Calculations show that this is a well conceptualized model of high accuracy and that the data extrapolation of the reservoir simulation model and linear programming model is better than that of artificial neural network （ANN）. This study sheds some light on the mechanism of reservoir operation and the results are useful for planning of annual operation of the cascade stations and also for other similar one-reservoir multi-stage cascade systems.

关 键 词：水文学 参数率定 算法耦合 SCE-UA算法 NSGAII算法 alt+8 

分 类 号：TV697.1[水利工程—水利水电工程]