论文部分内容阅读
提出了基于支持向量机(SVM)的边坡可靠度分析新算法。该方法采用均匀设计确定样本点,通过一定数量的确定性计算来训练SVM,拟合边坡的功能函数;采用一阶可靠度方法(FORM)和迭代算法优化SVM模型,获得可靠度指标和验算点信息;在SVM模型基础上进一步通过二阶可靠度方法(SORM)和蒙特卡罗模拟(MCS)计算边坡的失稳概率。以两个典型边坡为例,通过与其他方法比较,证明了该方法的准确性和高效性。结果表明:提出的在标准正态空间(U空间)中取样并构建SVM,在原始空间(X空间)中计算功能函数的算法,有效地解决了具有相关非正态分布变量的可靠度分析问题,并且可很容易扩展到SORM的计算。算例结果证明,该方法的精度高于FORM;而效率优于MCS。分析过程中,边坡安全系数计算和可靠度分析相互独立。因此,该方法既适用于具有显式功能函数的简单问题,也适用于需要软件计算安全系数的实际边坡问题。
A new algorithm of slope reliability analysis based on Support Vector Machine (SVM) is proposed. This method uses a uniform design to determine the sample points and trains the SVM by a certain amount of deterministic calculation to fit the slope’s function. The first-order reliability method (FORM) and iterative algorithm are used to optimize the SVM model to get the reliability index and checking Point information. Based on the SVM model, the slope instability probability is further calculated by the second-order reliability method (SORM) and Monte Carlo simulation (MCS). Taking two typical slopes as an example, the accuracy and efficiency of the method are proved by comparison with other methods. The results show that the proposed algorithm of sampling and constructing SVM in normal normal space (U space) and calculating the functional function in the original space (X space) can effectively solve the reliability analysis problem with correlated non-normal distribution variables , And can be easily extended to SORM calculations. The results show that the accuracy of this method is higher than that of FORM and the efficiency is better than that of MCS. During the analysis, slope safety factor calculation and reliability analysis are independent of each other. Therefore, this method is suitable for both simple problems with explicit function functions and real slopes that require software to calculate the safety factor.