论文部分内容阅读
本文首先介绍基于归一化的锥尖阻力Q_t和归一化的摩阻比F_R等触探参数的Robertson分类图及其解析表达形式,再根据研究区域的土层分布状况,简化Robertson分类图和贝叶斯模型。最后收集江西省某高速公路沿线溶洞上覆土层89个CPT和钻孔取样资料,分别采用最大似然法和贝叶斯法对Robertson分类图中的边界进行修正和比较,发现变异系数COV越小,先验分布越接近Robertson土壤分类图,预测结果越接近于Robertson土壤分类图,准确率也较以往的70%有所提高。变异系数COV越大,先验分布越为含糊不清,预测结果越接近于最大似然法的结果,但由于考虑了先验分布,准确率仍高于最大似然法。而在样本数量有限的情况下,最大似然法计算结果与Robertson分类图存在较大差别,准确率较差,应谨慎使用。
In this paper, the Robertson classification map based on the normalized cone tip resistance Q_t and the normalized friction resistance ratio F_R and other analytical expressions are introduced. Based on the distribution of soil layers in the study area, the Robertson classification map and Bayesian model. Finally, 89 CPT and sampling data of boreholes overlying an expressway in Jiangxi Province were collected. The maximum likelihood method and Bayesian method were used to correct and compare the Robertson classification maps. The results showed that the smaller the coefficient of variation COV , The closer the a priori distribution is to the Robertson soil classification map, the closer the prediction results are to the Robertson soil classification map, and the accuracy rate is also 70% higher than the previous one. The larger the coefficient of variation COV, the more the prior distribution is more ambiguous. The closer the prediction results are to the maximum likelihood method, the accuracy is still higher than the maximum likelihood method considering the prior distribution. In the case of a limited number of samples, the results of maximum likelihood method and the Robertson classification diagram are quite different, the accuracy is poor, should be used with caution.