摘要
基于自然风的特性,采用AR模型(自回归模型)模拟具有时间和空间相关性的随机风速时程,并对Whittle递推算法进行了改进,用于求解回归系数矩阵,编制了相应的风速时程模拟程序.算例表明:该法计算效率高,所编制程序可用于超大跨结构复杂屋面随机风场模拟.另外,通过对AR模型有关参数的分析讨论知:对于Davenport谱,兼顾精度和效率,时间步长取0.1s左右、截取频率上限取5~8Hz比较适宜.
Based on natural wind properties, AR model (Auto-Regressive Model) was employed to simulate wind speed time-history series, which had time and space correlativity. And the Whittle iteration method was modified, which was adopted to solve the matrix of regressive coefficients. In addition, the corresponding computational program was developed. Numerical examples have indicated that this method is feasible and highly efficient, and the program can be used in numerical simulation of stochastic wind field on super-long-span roof structures. Some parametric analyses on the AR model were carried out, which has shown that, for Davenport' s power spectrum, the preferential time step is about 0.1 s, and the optimum intercepting upper limit of the frequency should be taken as 5- 8 Hz.
出处
《湖南大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2008年第9期1-5,共5页
Journal of Hunan University:Natural Sciences
基金
国家自然科学基金资助项目(50508013)
湖南省杰出青年基金资助项目(08JJ1006)
教育部新世纪优秀人才资助项目(NCET-2007-0266)
关键词
数值模拟
超大跨屋面结构
风速时程
AR模型
相干函数
numerical simulation
super- long-span roof
wind speed time- history series
AR model
coherence function