考虑经验因素的暴雨频率曲线最优化拟合算法

Optimal fitting algorithm of rainstorm frequency curve considering the empirical factors

暴雨频率曲线拟合是推求暴雨强度公式必不可少的步骤,考虑经验因素进行暴雨频率曲线拟合,提出将暴雨强度频率曲线拟合作为最优化问题,采用加权阻尼高斯牛顿迭代算法求解。与已有方法相比,提出引入权重系数以提高工程常用重现期段拟合精度,避免不同历时暴雨频率曲线相交;提出应用有限差分法简化雅克比矩阵计算,并在海塞矩阵对角添加阻尼系数改进迭代收敛。以云南省保山市隆阳区33a实测降雨资料为例,证明了算法的可行性及实用性。

The rainstorm frequency curve fitting is essential for the identification of storm intensity formula,the study of rainstorm frequency curve fitting with considering of experience factors was carried out,and put forward to regard the rainstorm intensity frequency curve fitting as an optimization problem,and then to solve it by the weighted damped Gauss-Newton iterative algorithm.Compared to existing methods,the proposed method introduced weight coefficients to improve the fitting precision of commonly used recurrence period in engineering,and to avoid the intersection problem of different frequency curves.The finite difference method is proposed to simplify the calculation of Jacobian matrix,and the damping coefficient was added in Hesse matrix to improve iterative convergence.Thirty-three years of rainfall data of Longyang District of Baoshan city in Yunnan Province were used as an example to illustrate and demonstrate the feasibility and practicability of the proposed algorithm.