基于SCE-UA算法的API模型应用研究

Study on the Application of API Model Based on SCE-UA Algorithm

为了减小Nash模型参数优化中人工试错法和局部优化法的不确定性，以一种快速有效的优化方法求Nash模型参数的全局最优解。以安徽呈村和榆村流域为例，使用SCE—UA算法对API模型参数进行优化，采用对数绝对值误差为目标参数，分析优化结果。选用1988年～1999年共20场洪水进行经检验分析，模拟结果确定性系数均达到0．8以上．其中场18洪确定性系数达到0．9以上。研究结果表明，采用SCE—UA算法对API模型参数优化可以取得较好的结果。提高了模拟精度。

In order to find the global optimal parameters of the Nash model effectively and meanwhile to reduce the uncertainty caused by manual calibration and local optimization methods, two case studies are performed by applied SCE-UA algorithm to optimize the parameters of API Model in Chengcun and Yucun watersheds, Anhui Province. The calculated flow and measured discharge logarithm absolute errors are chosen as the objective functions of hourly simulation respectively to analyze optimized parameters. The optimized parameters which are verified by 20 floods between 1988 to 1999 indicate that the Nash-Suttcliffe coefficients in hourly simulation are equal or greater than 0.8, in which, the coefficients for 18 floods are equal or greater than 0.9. The results show that the SCE-UA algorithm is capable of finding global optimum of the API model and the accuracy and deterministic coefficient can be improved.