清江流域7月20日以后洪量明显减小，可以根据洪水和长江中下游出梅时间的预报结果考虑提前蓄水。建立梯级水库提前蓄水多目标联合优化调度模型，以旬最小生态流量作为蓄水期下泄流量的刚性约束，并采用NSGA-II和PA-DDS优化算法求解。应用结果表明：Pareto前沿分布范围均匀且广泛，可供决策者灵活调度。与原设计方案相比，通过水库群提前蓄水联合优化调度，总蓄满率由82%增加到89%，蓄水期可增发电量4100万kW•h (+2.1%)，增加生态供水1.1亿m3。 Since the flood volume in Qing River basin is decreased evidently after July 20, the reservoir early im-pound operation might be considered based on the forecasts of flood and out-time Meiyu in the middle and lower Yangtze River basin. A multiobjective early impound operation model for cascade reservoirs is established with ecologic flow constrains and optimized by the NSGA-II and PA-DDS algorithms. Application results demonstrate that the Pareto front has wide and uniform distribution which provides a flexible selection for decision-makers. Compared with the original impoundment operation rule, the optimal solution can enhance total impoundment efficiency from 82% to 89%, generate 41 million kW•h (or 2.1%) more hydropower annually and increase 0.11 billion m3 elological flow during impound period.