随机结构可靠度分析的支持向量机方法.pdfVIP

随机结构可靠度分析的支持向量机方法 随机结构可靠度分析的支持向量机方法 赵卫，刘济科 ( 中山大学应用力学与工程系, 广东 广州 510275) 摘 要：将拉丁超立方抽样法用于产生样本，用支持向量机方法逼近分布函数，提出了一种将拉丁超立方抽 样法和支持向量机方法相结合用于随机结构的可靠度分析的新方法。将本文方法与目前使用的一些随机结 构可靠度计算方法进行对比分析，阐明了本文方法的优点。对于一系列不同的极限状态函数条件下的随机 可靠度计算问题，分别采用本文方法和其它方法诸如 Monte-Carlo 和 FORM （一次可靠度方法）等进行了 计算比较。算例结果表明，本文方法不仅节省样本数量，而且相对于 Monte-Carlo 等方法大大降低了计算成 本，计算结果具有较高的精度。 关键词：随机结构；可靠度；支持向量机；超立方抽样. Support vector machine method for reliability analysis of stochastic structures ZHAO Wei ， LIU Ji-ke (Department of Mechanics, Zhongshan University, Guangzhou 510275, China) Abstract: This paper proposes a new reliability analysis method of stochastic structures by combining Latin hypercube sampling method with support vector machine. Latin hypercube sampling approach and support vector machine method are respectively used to generate the samples and estimate the distribution function. On the basis of analysis on current approaches of stochastic structure reliability calculating approaches, points out the advantages of the new method. For the problems in conditions of a series of different limit state functions, calculating and comparing are carried out respectively adopting the new method and other approaches of stochastic structure reliability determination such as Monte-Carlo method (MCM) and FORM (first order reliability method). Calculation results show that the present approach not only saves sample quantity but also greatly reduces the calculating cost compared with other methods such as Monte-Carlo and etc., and the results have higher precision. Key Words: stochasti