【摘要】 在长期调节水电站的水库优化调度中,一般采用马氏决策规划寻求平均年期望发电量达到最大的最优策略。但是,在极大多数情况下,这种策略往往并不能满足发电保证率的条件。 本文首先叙述了在相邻时段径流相关和面临时段径流不能予报的条件下,应用马氏决策规划的一般方法。然后,根据水电站保证率的含义,给出其概率形式的定义,从而把问题归结为随机约束下的马氏决策规划。 我们对应用惩罚方法求解这一问题进行了研究,对特定的罚函数得到了若干有用的性质。最后,简要地给出某水电站水库的一些实际计算结果。更多还原

【Abstract】 In the problems of optimization of a long-term regulating storage reservoir of a hydropower station, the Marcovian decision programming is often used to find the optimal policy of maximizing the average expected annual electric energy. However, in the majority of cases, it does not satisfy the reliability requirement of electric generation.In this paper, we first describe the general method of applying Marcovian decision programming under the conditions that the stream flow forecast is not available, and that there exists the correlation of stream flow between the coming time interval and its adjacent one. Then, according to the meaning of reliability of hydropower station, we give its definition in probability form. Thus, the problem can be reduced to a Marcovian decision programming, subjected to a stochastic constraint.An algorithm based on the penalty method is developed to solve such problems and some useful properties of specified penalty function are obtained.Finally, we present some numerical results of the calculation of optimization regulation of a hydropower station.更多还原