Abstract: High-precision optimization algorithm can improve the flood benefit of reservoir optimization operation. Aiming at the problem that the traditional progressive optimization algorithm (POA) is easy to fall into local optima, this study improves the POA algorithm, introduces the simulated annealing (SA) algorithm into the POA calculation framework, and proposes a SA-POA algorithm.This method discards the optimal solution of the optimization direction by a certain probability, so that the optimization result is less likely to fall into the local optima, thereby improving the quality of the solution. In order to further verify the practical effect of this method, taking "23·7" basin-wide extreme flood of Xidayang Reservoir of Tangxian County of Hebei Province as an example, the traditional POA, particle swarm optimization (PSO) and stepwise optimization-simulated annealing (SA-POA) algorithm are used to solve the problem and compare with the scheduling results. The results show that the peak clipping rate of SA-POA algorithm is 6.5% higher than that of POA algorithm under the maximum peak clipping criterion. Under the maximum water level minimization criterion, the maximum reservoir water level of the SA-POA algorithm is 0.5 meters lower than that of the POA algorithm. The SA-POA algorithm shows better solving performance under both criteria.