作　　者： ; ;

机构地区： 天津大学建筑工程学院

出　　处： 《哈尔滨工业大学学报》 2009年第4期169-172,共4页

摘　　要： 为了进行各种极限状态方程不明确的大型结构可靠度分析,提出了人工智能响应面法.利用BP网络来模拟响应面,基于遗传算法的优化法来进行可靠指标计算,确定性有限元分析采用ANSYS软件进行.综合VC与Matlab混合编程技术、基于VC和APDL的ANSYS二次开发技术及M atlab的遗传算法(GA)和神经网络(NN)工具箱函数,编制了可以适应各种复杂工程结构的可靠度分析程序.算例分析表明本文方法计算精度高.本文方法与程序适用于随机变量数目较多,功能函数为隐式时的复杂结构可靠度分析. In order to calculate the reliability of large scale civil structures without explicit performance functions, an artificial intelligence response surface method is proposed. The response surface was simulated by BP neural network, the genetic algorithm was applied to evaluate the reliability index, and the structure analysis was carried out by ANSYS. A program based on the proposed method was also developed by carrying out the mixed- language programming in Visual C ＋＋ 6. 0, integrating standard APDL （Ansys Parametric Design Language） and GA as well as NN toolboxes in Matlab language. The validity of present program, was verified by some numerical examples. The present method and program can be used to evaluate the reliability of complex structures with implicit performance functions.

关 键 词： 结构可靠度 神经网络 遗传算法 响应面法 程序开发

领　　域： [建筑科学] [自动化与计算机技术] [自动化与计算机技术]