地基沉降修正系数的Bayes概率推断

通过分析常规方法在沉降修正系数的选取中具有的定值性和随意性，引入建立在过去信息和现在样本信息之上的Bayes理论，结合某客运专线红黏土路基工程，提出用后验分布得到修正系数的取值范围。实例研究表明，用以往经验综合样本信息，估计修正系数的先验概率在某一区间上服从均匀分布。由现场载荷试验实测沉降量与理论计算沉降量分析所得的修正系数，将现场量测的沉降变形信息与先验信息结合起来，利用Bayes统计理论，由小样本试验数据推算得到修正系数的后验概率服从正态分布。对后验分布所得参数进行区间估计，得到该区域红黏土地基沉降修正系数的取值优化区间为 1.0, 1.7]，分析了不同荷载作用条件下沉降修正系数的概率分布模型。

Bayesian statistics conduct on modified coefficients for foundation settlement calculation

Through the analysis of the definite and randomness in choosing the settlement modified coefficients by the traditional calculation methods for foundation deformation, the Bayesian statistics was brought forward which was built on the old information and the sample information. The posterior distribution of settlement modified coefficients had been obtained. Conjoining the road bed engineering structure of one passenger dedicated railway, the value range of settlement modified coefficients had been obtained through its posterior distribution. The case study shows the priori probabilities of modified coefficients were estimated to obey uniform distribution in the certain range based on the old information. The settlement modified coefficients had been obtained through analyzing the observed deformation and calculation deformation. Combining the deformation information from the field tests with the prior information, the posterior probabilities of modified coefficients were calculated to obey normal distribution by Bayesian statistics. In addition, the parameters were investigated by estimation of Bayesian confidence interval. According to this red clay, interval estimation of modified coefficients was optimized at 1.0, 1.7]. And probability distribution models of modified coefficients had been obtained under different load conditions.