基于径向基函数响应面法的岩质边坡可靠度分析

岩质边坡中存在多种变异性,需要进行岩质边坡的可靠度分析。将数值分析法与可靠度分析的响应面法相结合,进行了岩质边坡稳定的可靠度分析。利用径向基函数网络(RBFN)的全局适应能力强的特点,在响应面法中,以径向基函数网络为响应面函数,将隐式的功能函数显式化。通过对每个响应面新增一个样本点的方法进行样本点的生成及响应面的迭代,计算表明:基于径向基函数网络的响应面法具有较高的计算精度及效率,可以适用于基本变量数较多的可靠度分析问题;本文的边坡算例中,第2层岩体的强度参数对边坡的可靠指标影响最大,而其他参数对可靠指标的影响则相对较小。

Reliability Analysis of Rock Slope Using RBFN-based Response Surface Method

There is much variability in rock slopes, so it is very important to carry out reliability analysis of rock slopes. The numerical analysis method of rock slopes and the response surface method of reliability analysis are combined together to perform the reliability analysis of rock slopes. Because the robust ability of global adaptability of the radial basis function network, it is used as the response surface function to explicit the implicit limit state function in the response surface method. A new sampling method which adds only one sample point for generating each new response surface is proposed to perform the generation of sample points and iteration of the response surfaces. It is demonstrated that the radial basis function network based response surface method (RBFN-based RSM) is accurate and efficient for calculating the reliability index. This method is very suitable for reliability analysis when the number of basic parameters is relatively large because it can greatly reduce the number of callings to the subroutine of numerical analysis of rock slopes. For the rock slope studied in this paper, the strength parameters of the second rock layer have a large influence on the reliability index of the rock slope while the other parameters’ influence on the reliability index is relatively small.