流体饱和多孔介质参数反演的遗传算法

GENETIC ALGORITHM FOR PARAMETER INVERSION OF SATURATED POROUS MEDIA

根据位移响应的理论合成应与实际测量数据相拟合的原则,将流体饱和多孔介质参数反演问题归结为非线性多峰函数的最优化问题。全局最优解的求解采用了遗传算法,克服了传统的优化法难于求得全局最优的困难。数值算例的实践表明了遗传算法的可行性和稳健性。

Fluid saturated porous medium (FSPM) is a two-phase medium composed of solid and fluid. Contrasting with classical single-phase elastic media, the two-phase medium is more close to actual earth stratum, hence, FSPM model is widely used in various engineering such as the geophysical prospecting, earthquake engineering, and rock and soil dynamics, etc. According to the principle that the computed response and measured response should be fitted. The parameter inversion problem of porous media is reduced to optimal problem of non-linear multimode function. The traditional optimal methods, such as gradient method, perturbation method, or time-convolution regularization iterative method, are based on Newton iterative method with local convergence. In order to overcome the difficulties met by classical optimal method, the genetic algorithm is used to search for the global optimal solution. At last, the genetic algorithm method is used to inverse the parameters of 2-D wave equations in porous media which has an analytical solution given by Paul in 1976. The numerical results confirm that genetic algorithm is feasible and robust.