研究并建立一种参数优选的改进遗传算法．该方法是一种隐并行非数值算法，它能较好地处理非线性、多峰、非凸的寻优问题．利用该算法对流域降雨－径流模型的九个参数进行优化，能克服传统优化方法把降雨径流物理过程割裂开来的缺陷；同时可求出产汇流过程参数群的值，实现整体优化．计算结果表明，该算法效果令人满意． Research and establish a parameter optimization of the improved genetic algorithm. This method is a kind of hidden parallel non-numerical algorithm, which can better deal with the nonlinear, multi-peak, non-convex optimization problems. Using this algorithm to optimize nine parameters of rainfall-runoff model can overcome the defect that the traditional optimization method breaks the physical process of rainfall-runoff. At the same time, we can find out the value of process parameter group of production flow and achieve the overall optimization. The calculation results show that the algorithm is satisfactory.