基于MATLAB的支持向量机结构可靠度分析方法  

Structural Reliability Analysis Using Supported Vector Machine and MATLAB

作　　者：陈铁冰[1] 邢媛媛[2] 谭也平[2] 

机构地区：[1]深圳职业技术学院建筑与环境工程学院,广东深圳518055 [2]深圳大学土木工程学院,广东深圳518060

出　　处：《华中科技大学学报（城市科学版）》2009年第1期60-63,共4页Journal of Huazhong University of Science and Technology

基　　金：广东省自然科学基金(06028131);深圳职业技术学院重点科技项目(06KJd007)

摘　　要：针对复杂结构极限状态方程一般难以显式表达的特点,提出了基于MATLAB的支持向量机结构可靠度分析方法。该方法利用MATLAB计算平台,采用均匀抽样法对随机变量进行抽样,形成结构响应与随机变量的样本值。通过对支持向量机进行训练,利用其高度非线性映射能力,模拟结构极限状态方程,应用几何法计算结构可靠指标。研究表明,采用支持向量机能够很好地拟合结构极限状态方程,计算结构可靠指标具有很高的精度。An approach evaluating the structural reliability using Supported Vector Machine （SVM） based on MATLAB is proposed in this paper when implicit limit state functions are normally encountered in the complicated structures. Random variables are sampled by average sampling method. SVM can be trained using the input-output pairs which are generated using samples from structural responses and random variables. The trained SVM can map the structural responses and random variables and the limit state functions of structures can be approximated using SVM. Then first order second moment reliability method（FORM） is used to calculate the reliability index of structures. Numerical examples are given. The results show that SVM can well approximate the limit state functions of structures. The accuracy and efficiency of the proposed approach in calculating the reliability index of structures are validated.

关 键 词：结构可靠度 MATLAB 支持向量机 结构极限状态方程 

分 类 号：TU311.2[建筑科学—结构工程]