水文学中雨强公式参数求解的一种最优化方法

提出了一种客观的、最优化的暴雨强度公式参数估算方法:先将公式线性化, 确定出未知参数b, C取值范围, 给定一个b值 (分公式)、b, C组合 (总公式), 再对雨强-历时-重现期 (i-t-T) 三联表数据进行最小二乘法拟合可得到参数A, n, 以总误差最小为控制条件, 理论上可得到最优的一组参数估算值。并以深圳、武汉两市为例, 进行暴雨强度公式参数估算, 精度高于国家标准要求, 且明显优于对比方法。该法已被编制成计算机软件, 只要输入原始资料就可以很快输出结果, 包括曲线型估计、参数估算、误差分析、图表, 使用极其方便, 可向全国各地推广应用。

An Optimized Method for Estimating Parameters of the Rainstorm Intensity Formula

In hydrology,the form of the rainstorm intensity formula is known as the nonlinear one with excess factors,and its parameters are not easy to be solved with normal methods,which makes the method design of parameters estimate and the elimination of the error from parameters estimate the most crucial.For a long time, many methods to estimate the parameters of the formula have been put forward by hydrological and meteorological experts that have exerted a profound influence on urban water drainage design.But all the methods above have certain error from objective error by personal judging,approximate supposed error,or missing error from skip-over searching.An objective and optimized method for estimating the parameters of the rainstorm intensity formula is put forward that can avoid the above errors skillfully.First,the non-linear formula is linearized and the scope of the parameters b and C is decided.Second,the parameters A and n can be determined based on known rainstorm intensity,duration and return period(i-t-T) tri-relation table by least regression method after the value of b(single period formula) or a combination of the value of b and C(multi-period formula) is given.Then the formula with least error is the optimum one.This method is used in Shenzhen and Wuhan separately,and the accuracy of the formula can meet the requirement of the national standard and is superior to the compared methods.A software has been designed according to the new method and can be used easily and popularized nationwide.If the original data are put in,the estimating result of the curve pattern,formula parameters,error,figures and table can be obtained quickly.