基于支持向量机和贝叶斯方法的岩体参数反分析

Back analysis of rock mass parameters based on support vector machine and Bayesian method

为实现岩土体参数随施工过程的动态不确定性反分析,将多输出支持向量机方法和贝叶斯理论引入岩体参数不确定性动态反分析中。以白鹤滩水电站左岸边坡为例,建立了基于位移与松驰深度的岩体参数概率反分析方法。研究结果表明:概率反分析方法提供了更多的关于岩土力学参数的信息,能更进一步地匹配地质工程的复杂性和不确定性特征,可以实现参数及其不确定性特征随施工过程的动态更新。所提出的方法具有可行性,可以推广应用到类似工程中。

In order to realize the dynamic uncertainty back analysis of rock and soil parameters along with the construction process, the multi output support vector machine method and Bayesian theory were introduced into the dynamic uncertainty back analysis of rock mass parameters.Aiming at the left bank slope of Baihetan Hydropower Station, we established a probability back analysis method of jointed rock mass parameters based on displacement and relaxation depth.The results showed that the proposed uncertainty back analysis method provided more information about geotechnical mechanical parameters, thus can further match the complexity and uncertainty characteristics of geological engineering, realizing dynamic updating of parameters and their uncertainty characteristics during the construction process.The proposed method is feasible and can be applied to similar projects.