中文摘要：

梯级水电站联合优化调度是一项涉及学科门类广泛、牵涉部门利益众多的复杂大系统优化决策问题,对制定和实施区域用水规划、实现经济社会可持续发展具有重大的现实意义。鉴于当前群体智能优化算法应用于梯级水电站联合优化调度中存在的＂维数灾＂及大量约束条件不易处理的难点,将加速遗传算法（AGA）应用于梯级水电站联合优化调度研究中,采用＂分类假设＂的思路逆序寻找不同电站、不同时段优化变量可行决策空间并生成初始种群个体,由此重点阐述了改进遗传算法对优化调度模型大量复杂约束条件的实现方法。上述方法在我国水、电特性代表性良好的乌江梯级七库联合优化调度实例的应用结果表明：加速遗传算法对梯级水电站联合优化调度模型复杂约束条件具有较强的自适应及全局搜索能力,且计算结果与设计成果相比,乌江梯级水电站多年平均发电量增加约2.60%。可见,采用＂分类假设＂的研究思路处理群体智能优化算法应用于梯级水电站联合优化调度中存在的复杂约束问题是合理可行的,可为流域梯级水电站实行集中运行、调度提供科学有效的决策依据。

英文摘要：

Combined optimal operation of cascade hydropower stations is an optimized decision-making problem of large complicated system,and it involves various disciplines and agencies.Study of this issue is of great significance to regional water-use planning and sustainable development of regional economy.When a swarm intelligent optimization algorithm is applied to this problem,a large system of complex constraint conditions and curse of dimensionality often cause practical difficulties in their treatments.In this study,an accelerating genetic algorithm（AGA） was adopted to solve the problem.By principle of classification and hypothesis,a feasible decision space with a reverse time was constructed for the optimization variables of different stations,and then initial population individuals could be generated.This paper focuses on a discussion of the approaches for implementing and realizing this constraint system by AGA.Application to Wujiang cascade hydropower stations of typical hydraulic and power features shows that AGA has a strong adaptability and global searching ability of the constraint system.In comparison with the design case,the average annual yield of the stations is increased by 2.60%.Thus use of the swarm intelligent optimization algorithm to optimal operation problem by the principle above is feasible and reasonable,and this work provides a tool of scientific and effective decision for centralized dispatching operation of cascade hydropower stations.