摘要
为了对混凝土结构进行收缩徐变效应随机分析,提出一种复合算法。该算法利用拉丁超立方抽样技术对随机变量进行抽样,通过有限元逐步分析得到收缩徐变效应下的结构响应量,建立显式化神经网络响应面,在此基础上进行拉丁超立方抽样蒙特卡洛数值仿真和参数敏感性随机分析。算例表明:所提算法能够兼顾精度和效率,便于程序实现,能为复杂结构收缩徐变效应随机分析提供有效工具。
In order to carry out stochastic analysis of concrete structures due to shrinkage and creep effect,a new hybrid algorithm was proposed. This algorithm used the latin hypercube sampling technique to select the random variables sample,and the structural response of the shrinkage and creep effect was obtained by the finite element gradual analysis,and then a response surface of neural network was set up,based on which Monte Carlo simulation of latin hypercube sampling and random analysis of parameter sensibility were carried out. Numerical examples showed that the proposed algorithm could provide effective tools with accuracy and efficiency for the stochastic analysis of complex structures due to shrinkage and creep effects.
出处
《工业建筑》
CSCD
北大核心
2016年第9期76-80,55,共6页
Industrial Construction
基金
高等学校博士学科点专项科研基金(20133204120015)
江苏省高等院校自然基金(12KJB560003)
国家杰出青年科学基金项目(50725828)
国家自然科学基金项目(51278104)
关键词
混凝土
收缩
徐变
随机分析
concrete
shrinkage
creep
stochastic analysis
