基于FEM-SVM的大坝可靠度分析方法

为解决复杂大坝可靠度分析计算时难以求得显示的功能函数和传统随机有限元计算时间长、工作量大等问题,提出将有限元(FEM)、支持向量机(SVM)和Monte-Carlo法相结合的可靠度分析方法,即采用正交试验设计具有代表性的样本点,通过有限元计算生成学习样本,利用支持向量机具有解决高度非线性、小样本问题的能力,建立结构响应量和输入变量的映射关系,再利用Monte-Carlo法,结合输入变量的概率分布生成随机样本进行结构可靠度计算。实例计算表明,FEM-SVM分析方法具有较好的精度,同时大幅减少了计算时长和工作量。

Dam Reliability Analysis Method Based on FEM SVM

Aiming at the reliability analysis problems with implicit performance functions of complex dam and reducing the computation time and workload, a reliability analysis method combined with the FEM, support vector machine (SVM) and Monte Carlo was presented. The orthogonal experimental method was used to design the typical samples. The learning samples were generated by the finite element calculating. Then support vector machine which has the ability to solve the problem of highly nonlinearity and small sample was adopted to establish the structural response and the mapping relation of input variables. Combined with the probability distribution of the input variables, random samples were generated by using Monte Carlo method. Finally, the structural reliability was calculated. An engineering example results show that the accuracy was improved and the computing time and effort were reduced by using the proposed method.