土体渗流固结参数识别方法

Estimation of soil parameters under consolidation based genetic algorithm

根据土体固结过程中超孔隙水压力观测资料 ,建立了基于遗传算法的土体渗流固结参数非线性识别方法 ,解决了经典高斯 牛顿极小化问题所存在的局部极小问题和最小二乘法所存在的当初始值选择不合适时迭代过程发散的问题 ,提出了根据观测仪器的精度 ,建立迭代终止条件的方法。数值计算结果表明 。

According to the data of pore water pressure monitored in the field under consolidation, the non\|linear estimation procedure of soil parameters was researched based on the genetic algorithm in the paper. It was successfully overcome that there were some problems of the local minima and non\|convergence in the classic Gauss\|Newton minimization and the least square method because the genetic algorithm is global convergent, to usually achieve computational efficiency, and to have some level of robustness against entrapment in local minima. The premature convergence problem in the GA was also avoided through adjusting to fitting function. The termination criteria of iteration was firstly proposed according to the observing precision of the equipment. The computational results fact that the non\|linear inversion procedure proposed is global convergent, to have faster computational efficiency, and stronger stability.