论文部分内容阅读
根据基坑开挖过程中获取的现场监测数据可以反分析现场土层的土体参数.贝叶斯方法为充分利用现场监测数据分析土体参数提供了一条有效的途径.实现贝叶斯更新一般有两种方法:一是各开挖阶段的观测值依次对土体参数进行更新,即分步更新,更新过程中需假设后验分布类型,如正态分布;二是将多阶段的观测值作为整体对土体参数进行更新,即整体更新.为了研究贝叶斯更新方法对更新结果的影响,在先验分布为正态或均匀分布时分别采用这两种方法对台北国营企业中心开挖工程的土体参数进行更新.结果表明:当先验分布为正态分布时,假设后验分布为正态分布进行分步更新,得到的参数分布基本合理,能够准确预测开挖引起的最大沉降和失效概率;相反,当先验分布为均匀分布时,假设后验分布为正态分布进行分步更新,不能够准确预测开挖引起的最大沉降和失效概率,此时应采用整体更新方法.
According to the on-site monitoring data obtained during excavation, the soil parameters can be back analyzed.The Bayesian method provides an effective way to fully utilize the on-site monitoring data to analyze soil parameters. There are two methods: First, the observations of each stage of excavation in turn update the soil parameters, that is, step by step update, the update process needs to assume the posterior distribution types, such as normal distribution; Second, the multi-stage observations As a whole to update the soil parameters, that is, the overall update.In order to study the impact of Bayesian update method on the update results, when the prior distribution is normal or uniform distribution, these two methods were used to excavate the state-owned enterprise center in Taipei Engineering soil parameters are updated.The results show that when the prior distribution is normal distribution, the posterior distribution is assumed to be a normal distribution step by step update, the parameters obtained are basically reasonable distribution, can accurately predict the maximum settlement caused by excavation And failure probability. On the contrary, when the prior distribution is uniform distribution, it is assumed that the posterior distribution is updated step by step as a normal distribution, and the maximum settlement and loss caused by excavation can not be predicted accurately Effective probability, this time should be used as a whole update method.