以新安江三水源模型为例，分析了在不同土壤前期含水量、不同降雨强度和雨型、以及不同暴雨中心位置等情况下模型参数的特征规律。在此基础上重点探讨了模型参数智能化的方法和途径。最后以浙江曹娥江上游长诏小流域为实验区进行了分析验证。模拟结果表明该方法可较好地提高模型预报精度，且为洪水预报系统的智能化奠定了基础。 Taking the Xin’anjiang three water source model as an example, the characteristics of the model parameters in different soil pre-stage water content, rainfall intensity and rain type, and different storm center locations were analyzed. On this basis, the methods and ways of intelligent parameterization of the model are discussed emphatically. Finally, the Changzhao catchment in the upstream of Cao’e River in Zhejiang Province was used as an experimental area for verification. The simulation results show that this method can improve the accuracy of model prediction and lay the foundation for the intelligence of flood forecasting system.