中文摘要：

由于未考虑时段之间的关联关系，现行的采用回归分析提取调度函数的方法在模拟长系列径流计算时运行效益较差且发电保证率不高。以狮子滩水库为例，在建立以水电站总发电量最大为目标函数的优化调度模型基础上，用2元线性回归提取各时段独立的调度函数系数作为初始值，采用改进的基于实数编码的加速遗传算法（RAGA）优化各时段的调度函数系数，考虑时段间的关联关系，再应用该调度函数模拟长系列径流输入到水电站的发电调度过程。计算结果表明，水电站的运行效益更好，而且发电保证率能满足要求。

英文摘要：

The relationship between the periods is not considered, thus the method used to extract the dispatching function by binary linear regression at present gets poor running benefits, as well as low guarantee rate. A case study of the Shizitan Hydropower Sta- tion, the monthly independent coefficients of dispatching function are extracted by binary linear regression as the initial value, based on the optimal operation model is established which aims at the maximum operational benefits. These initial coefficients are optimized by improved real coding based accelerating genetic algorithm and used to simulate the generating operation of long series runoff. We consider those relationships at the same time. The calculations show that the operational benefits are better and the guarantee rate o{ hydropower stations can meet the requirement.