针对地下水渗流随机参数反演问题,在正演随机模拟方法的基础上,结合非平稳随机过程变量的一阶Taylor展开式和随机参数的摄动方程,讨论了多随机参数反演问题,提出了基于抽水试验的渗流问题摄动随机反演方法,给出一阶均值反演准则,得到多随机变量的均值和方差的近似表示方法。该方法在计算均值时采用改进的遗传算法,在计算方差时采用分组统计的方法。最后给出了一个基于Thies模型反演导水系数和贮水系数的例子,计算结果表明,该方法可以较好的求取参数的均值,近似地求取参数方差。
The non-stationary stochastic variables are expanded by Taylor series and a perturbation equation is established for determining the multiple parameters in inverse analysis of seepage problem in pump test. A criterion for calculating the first order average value of the parameters is suggested and the method for determining the approximate average value and variance of multiple parameters is proposed. An improved genetic algorithm is adopted for calculating the mean values and the statistical method is applied to calculate the variances. The comparison of the stochastic hydraulic conductivity and storage coefficient identification of an example using the proposed method with Thies model shows that the proposed method gives good mean values and approximate variance of the parameters.