论文部分内容阅读
在考虑多个土体参数空间变异性的基础上,提出了基于拉丁超立方抽样的非饱和土坡稳定可靠度分析的非侵入式随机有限元法。利用Hermite随机多项式展开拟合边坡安全系数与输入参数间的隐式函数关系,采用拉丁超立方抽样技术产生输入参数样本点,通过Karhunen-Loève展开方法离散土体渗透系数、有效黏聚力和内摩擦角随机场,并编写了计算程序NISFEM-KL-LHS。研究了该方法在稳定渗流条件下非饱和土坡可靠度分析中的应用。结果表明:非侵入式随机有限元法为考虑多个土体参数空间变异性的非饱和土坡可靠度问题提供了一种有效的分析工具。土体渗透系数空间变异性和坡面降雨强度对边坡地下水位和最危险滑动面位置均有明显的影响。当降雨强度与饱和渗透系数的比值大于0.01时,边坡失效概率急剧增加。当土体参数变异性或者参数间负相关性较大时,忽略土体参数空间变异性会明显高估边坡失效概率。
Based on the spatial heterogeneity of soil parameters, a non-intrusive stochastic finite element method based on Latin hypercube sampling is proposed to analyze the stability of unsaturated soil slopes. Hermite random polynomials are used to develop fitting implicit function relationship between slope safety factor and input parameters. Latin hypercube sampling technique is used to generate input parameter sample points. The Karhunen-Loève expansion method is used to discretize the soil permeability coefficient, effective cohesion and Internal friction angle with the random field, and write the calculation program NISFEM-KL-LHS. The application of this method to the reliability analysis of unsaturated soil slopes under steady seepage conditions is studied. The results show that the non-invasive stochastic finite element method provides an effective analytical tool for the reliability problem of unsaturated soil slopes considering the spatial variability of soil parameters. Spatial variability of soil permeability coefficient and slope rainfall intensity have obvious influence on slope groundwater level and the position of the most dangerous slip surface. When the ratio of rainfall intensity to saturated permeability coefficient is greater than 0.01, the failure probability of slope increases sharply. When the variability of soil parameters or the negative correlation between parameters is large, ignoring the spatial variability of soil parameters will obviously overestimate the failure probability of slope.