Research on the optimization of hydrological model parameters is an important issue in the field of hydrological forecasts, as these parameters not only directly impact the accuracy of forecast programs, but also relate to the development, application, and popularization of hydrological models. In this paper we selected the double-excess runoff generation model as the subject for research, and the data obtained from tens of flooding events in the Fen River Basin were used for the construction of these models. The SCE-UA and MOSCDE algorithms were then taken to optimize the models’ parameters. The results showed that: as compared with the SCE-UA algorithm, higher flood forecast accuracies were obtained through model parameter optimization using the MOSCDE algorithm. During the examination period, the compliance rate of the flood peak magnitude increased from 60% to 70%, while the compliance rate of the flood peak duration increased from 80% to 90%. The Nash-Sutcliffe efficiency (NSE) of the flood peak magnitudes increased from 0.664 to 0.878, which demonstrates an improvement in goodness-of-fit; the RMSE value of flood peak magnitudes also decreased from 399.8 to 236.84, thus showing a decrease in dispersion and an improvement in goodness-of-fit. With the continuous improvements made in hydrological parameter algorithms and the creation of new optimization algorithms, there is no doubt that the optimization of hydrological model parameters will become more reasonable.