遗传算法和高斯牛顿法联合反演地下水渗流模型参数  

Parameters identification of groundwater flow model with genetic algorithm and Gauss-Newton Method

作　　者：姚磊华[1] 

机构地区：[1]中国地质大学工程技术学院,北京100083

出　　处：《岩土工程学报》2005年第8期885-890,共6页Chinese Journal of Geotechnical Engineering

基　　金：国家自然科学基金资助项目(40372110)

摘　　要：充分利用遗传算法善于进行全局搜索和高斯牛顿法善于进行局部搜索的优点,克服了两种方法各自的不足,用改进的遗传算法和高斯牛顿法联合反演地下水数值模型参数,首先用遗传算法求出地下水模型参数的初值,然后利用这组初值用高斯牛顿法进行数值模型参数的反演,并以非均质各向同性承压二维非稳定流动模型,结合有限元法讨论了用遗传算法和高斯牛顿法联合反演地下水数值模型参数的过程。计算结果表明,联合参数反演方法,具有收敛速度快、解的精度高的特点。With genetic algorithm （GA） regard to nonlinear optimization problems, has a good capability of searching in the whole solving space. But in the part of solving space, GA has a low capability of searching and gives a low precision of solution. As for Gauss-Newton Method （GNM） nonlinear optimization problems, has inverse characters opposite from GA. The strongpoint of GA and GNM were full used in parameters identification of ground water flow numerical model. At first, GA solved the initial values of parameter. Then, the parameters were identified by GNM with the initial values of parameter. A 2-Dimensional unsteady flows in an inhomogeneous, isotropic confined aquifer for an ideal model was employed, and the application of GA and GNM to inverse problems of hydrogeology parameters with finite element method was discussed. It is shown by the calculated results that the algorithm has several main functions, such as expediting constringency, and improving precision.

关 键 词：遗传算法 高斯牛顿法 反演 地下水数值模型 参数 

分 类 号：TU473.12[建筑科学—结构工程]