针对梯级水库群联合防洪优化调度问题，提出一种基于自适应成功历史策略的改进差分进化算法（strategy and parameter adaptive differential evolution，SPADE）。该算法通过自适应成功历史差分策略来提升随机搜索效率，通过精英种群保守策略提升局部收敛速度及全局探索能力。据此开展包含10个测试函数的数值实验和赣江中游梯级水库群联合防洪优化调度实例，用于检验所提出的算法应用效果。结果表明：在数值实验中，SPADE算法收敛结果的最优值、平均值、标准差和成功次数评价指标整体优于SHADE、自适应差分进化算法(self-adaptive differential evolution，SADE)、遗传算法（genetic algorithm，GA）、粒子群算法（particle swarm optimization，PSO）、人工蜂群算法（artificial bee colony，ABC）；在梯级水库群联合防洪优化调度实例应用中，通过对1964单峰和1973多峰型历史洪水过程进行分析，发现SPADE算法结果在削峰率指标上明显优于SADE、GA、PSO算法，且相比SHADE在两次历史洪水条件下的削峰率指标结果分别提升0.9%、3.4%。实验结果充分验证所提SPADE算法的优越性，可作为梯级水库群联合优化调度问题的有效求解工具。

Joint flood control and optimal scheduling of the cascade reservoirs is a mathematical problem that transforms the complex reservoir scheduling problems in the actual process into abstract optimization problems in the mathematical sense, and transforms these problems into different constraints, and seeks the optimal solution based on these constraints. The combined operation of reservoir groups needs to consider the influence of meteorological, hydrological, hydraulic and other factors, as well as the conflict of interests between upstream and downstream and between multi-functions. Domestic and foreign scholars have used dynamic programming (DP), progressive optimization algorithm (POA), genetic algorithm (GA), and particle swarm algorithm (PSO) and other algorithms to solve the problem. However, the above traditional optimization algorithms still have the problems of poor stability and easy to fall into the local optimal situation, and still need to carry out more in-depth research on the algorithm parameter updating mechanism, search strategy and other aspects.SPADE algorithm is an improved differential evolution algorithm, which uses adaptive success history difference strategy to improve random search efficiency, and adaptive success history parameter update strategy and elite group conservative strategy to improve local convergence speed and global search capability. The algorithm divides all the populations in each generation into elite and base populations in the process of differential variation, the elite population conserves the good genes without adopting the variation strategy, and individuals in the base population randomly select each variation strategy for evolution according to the probability, in which the probability of selecting the differential strategy in each generation is reassigned according to the evolution success rate of each individual produced by the differential strategy, and the flow of the high-quality genes is effectively controlled. A joint flood control optimal scheduling model of a group of terrace reservoirs with the objective function of minimizing the maximum flow rate discharged from the reservoirs is established, combined with the flow constraints and their constraint violation evaluation indexes, and solved by applying the spade algorithm in order to improve the computational efficiency and the flood control and peak shaving ability.The effectiveness of the proposed algorithm is examined through numerical experiments containing 10 test functions and an example experiment of joint flood control optimal scheduling for the middle reaches of Ganjiang River cascade reservoir group. The relevant experimental results show that: In the numerical experiments, the evaluation indexes of the optimal value of the statistical convergence error, the average value, the standard deviation and the number of successes, the evaluation indexes of SPADE algorithm are better than those of SHADE, SADE, GA, PSO and ABC algorithms in eight test functions, and the performance is also more outstanding in the remaining two test functions; In the case experiment of joint flood control optimization of cascade reservoir group in the middle reaches of Ganjiang River with the goal of minimum sum of square discharge of Xiajiang Reservoir, SPADE is obviously superior to SADE, GA and PSO in the index of peak cutting rate by analyzing the historical flood process of 1964 single peak and 1973 multi-peak type. Compared with the two historical flood conditions, SHADE increased by 0.9% and 3.4% respectively, which reduced the peak flood discharge of Ganjiang River. The above analysis fully verifies the superiority of the proposed algorithm, which can be used as an effective tool to solve the cascade reservoir group joint scheduling problem.The above analysis fully verifies the superiority of an improved differential evolutionary algorithm based on the successful history adaptive strategy proposed and its effectiveness in the application of the joint scheduling problem of a group of terrace reservoirs. The combined flood control optimization operation of cascade reservoir groups can rationally utilize the flood control capacity of reservoir groups and effectively play the role of reservoir groups in flood blocking and peak cutting, which is of great significance to reduce the downstream flood control risk.