流体饱和孔隙介质参数反演的模拟退火算法

本文研究了模拟退火算法在流体饱和孔隙介质参数反演中的应用.通过计算响应数据与实测响应数据的拟合将参数反问题归结为最优化问题.由于流体饱和孔隙介质运动方程的复杂性,动力响应与材料参数之间呈复杂的非线性关系,优化目标函数是非凸多峰函数.传统的梯度类优化方法一方面受局部极值的困扰难以搜索到全局最优解; 另一方面确定搜索方向须进行复杂的参数敏度分析.为克服这些困难,本文应用模拟退火算法进行了多参数反演数值模拟,模拟结果表明了模拟退火算法的可行性和稳健性.

INVERSE PARAMETER ESTIMATION OF FLUID-SATURATED POROUS MEDIA BY SIMULATED ANNEALING ALGORITHMS

The simulated annealing algorithm is used to determine material parameters of fluid-saturated porous media from surface dynamic response data. Based on fitting of calculated results with experimental data, the inverse parameter estimation problem is transformed into an optimization problem. The complex motion equations and the nonlinear relations between dynamic response and material parameters result in a nonlinear and multimode objective function. The conventional gradient-orient optimal methods not only need complex sensitivity analysis of parameters to determine the search direction, but also fail usually to find the global extremum due to the pitfall of local extremum. In order to overcome the shortcoming, the simulated annealing algorithm is used. Numerical results indicate that the simulated annealing algorithm is robust for the inverse estimation problem of material parameters .