基于可靠度敏感性的神经网络法研究边坡稳定性

针对边坡工程结构功能函数不能显式表达的可靠性分析问题和非线性问题计算量大的弊端．研究结构可靠度敏感性，提出参数的相对敏感性分析方法，并基于该方法提出了神经网络法分析边坡稳定性。具体思路：由可靠指标对随机变量分布参数的相对敏感性分析，确定边坡可靠度主要影响参数；用神经网络模型近似替代响应量与基本变量间的隐式极限状态函数，根据蒙特卡罗模拟法，对网络模型进行可靠度分析，求解结构可靠度指标。基于可靠度敏感性的神经网络法．对均值和成层边坡进行稳定性分析，与传统可靠度计算方法相比．结果表明：该方法分析边坡稳定性是准确的且具有较高的计算效率。

Research on Slope Stability Based on Neural Networks Method of Reliability Sensitivity

A reliability analysis of slope stability using neural networks method is presented in this paper,which based on relative sensitivity analysis of the structural reliability. It is especially useful in such reliability analysis problems as slope stability whose performance functions are implicit and nonlinear problems have large amount of calculations. Specific ideas： To access the reliabili- ty of the main effects of slope parameters based on reliable indicator of the random variable distribution parameters of the relative sensitivity,with neural networks models approximate amount of alternative responses and basic variables of the implicit limit state functions,to analyze the reliability of network model and to solve structural reliability index based on the Monte Carlo simulation. Research on slope stability based on neural networks method of reliability sensitivity and compared with traditional reliability calculation methods,the results show that： the method of slope stability analysis is accurate and has higher efficiency.