地基压缩模量及沉降经验系数反演分析

BACK ANALYSIS BASED ON ACE NON-PARAMETRIC REGRESSION TO FOUNDATION COMPRESSION MODULUS

以分层总和法为地基沉降计算模型,基于各土层压缩模量和沉降经验系数的取值区间,利用均匀试验和ACE非参数回归技术构建土层压缩模量和沉降经验系数与地基沉降量之间的响应面关系;然后利用Monte Carlo模拟技术生成多组随机样本,并利用已建立的响应面模型预测随机样本对应的地基沉降量;最后利用现场监测数据作为约束条件,筛选出在约束条件内的沉降预测值所对应的数据样本,并将所筛选出的压缩模量作为土层真实压缩模量值的近似。将本文方法用于实际工程反演分析,充分验证了该方法的有效性。

In this paper, firstly, the layer-wise summation method is used as calculation model, based on the suitable value ranges of settlement experience coefficient and compression modulus, using uniform experiment and ACE non-parametric regression technical, the respond surface between compression modulus and settlement experience coefficient and foundation settlement is built; then, using Monte Carlo simulation technical produced large number of random samples and put them into responding surface one by one to predict its foundation settlement; at last, using the field monitor data as constraint condition, screen out the samples, which its foundation settlement predict value is satisfied constraint condition, and regard them as approximate value of real value. An engineering practice is used as example to back analysis; the feasibility of this method is verified.