边坡稳定性影响因素敏感性人工神经网络分析

传统的边坡稳定性影响因素敏感性分析基本上都是基于各参数变化的大量试算分析，这样不仅计算量大，而且数据准各工作复杂。基于人工神经网络的非线性映射功能，用RBF神经网络模型进行边坡稳定性影响因素敏感性分析，把正交表试验设计理论、效用函数理论与神经网络结合起来进行边坡影响因素敏感性分析，给出了敏感性分析的具体算法并以圆弧破坏边坡为例进行了分析，并将分析结果与用传统理论分析所得结果进行比较，结果显示，用人工神经网络方法进行边坡稳定性影响因素敏感性分析不仅可靠，而且方便简单。

SENSIBILITY ANALYSIS OF SLOPE STABILITY BASED ON ARTIFICIAL NEURAL NETWORK

The traditional sensibility analyses of the affecting factors for slope stability are based on plenty of trial calculation according to the variation of different parameters. These methods need not only plenty of calculation,but also complicated data preparation. A new method of the sensibility analysis of affecting factors for slope stability is proposed,using RBF artificial neural network (ANN) model according to nonlinear mapping function of ANN. Combining the perpendicularity list with theory of utility function and ANN,the assessment model is set up to achieve sensibility analysis. Compared with the results obtained by traditional limit balance analysis,the result by proposed method is of validity and convenience.