基于LSSVM与MCS的路基沉降可靠度分析

提出了一种计算路基沉降可靠度的新方法。基于FLAC中的修正剑桥模型，以最小二乘支持向量机为核心技术，结合蒙特卡罗法构建计算模型。由于修正剑桥模型参数较多，对模型参数进行了敏感性分析，将对沉降影响较大的参数确定为随机变量。选取训练样本对支持向量机进行训练，按照随机变量的概率分布进行抽样，馈送到最小二乘支持向量机得到相应的响应值，用Matlab编制程序完成可靠度计算，并进行了算例分析。计算结果表明，蒙特卡罗法结合支持向量机的沉降可靠度计算方法应用于公路软基沉降可靠度计算是可行的。

Reliability analysis of soft foundations settlement based on LSSVM and MCS

A new method for calculating foundation settlement reliability is proposed. The kernel technique is based on modified Cam-Clay model in FLAC and least squares support vector machine (LSSVM) combined with MCS. The sensitivity of parameters involved in modified Cam-Clay model is analyzed; and the sensitive parameters are chosen as random variables. The stochastic variables are sampled according to the probabilistic distribution. A program is programmed to calculate the reliability by Matlab. The results show that the method of MCS combined with LSSVM is worth to study for reliability analysis of soft foundations settlement.