基于改进的线性滤波自回归模型法的大跨空间结构风荷载模拟

Numerical Simulation of Wind Loads Acting on Long-span Spatial Structures Based on Modified Linear Filter Auto-regressive Model Method

为了准确模拟大跨空间结构的风荷载,提出了改进的线性滤波自回归模型(Auto-Regressive model)法。该方法基于传统线性滤波自回归模型,通过高斯-赛德尔(Gauss-Seidel)迭代法求解大跨空间结构风荷载模拟中的回归系数矩阵,进而对大跨空间结构的风荷载进行模拟。结果表明:改进的线性滤波AR模型法解决了自由度过大导致的回归系数矩阵不正定的问题;能够有效地模拟具有空间相关性、时间相关性的节点脉动风速时程,模拟精度、速度和计算稳定性均满足实际工程应用要求。

In order to simulate the wind loads of long-span spatial structures accurately,a modified linear filter Auto-Regressive model method was presented.The method,which is based on traditional linear filter auto-regressive model,solves the regressive coefficients matrix through Gauss-Seidel iteration method,and then simulates the wind loads of long-span spatial structures.It is found that modified linear filter auto-regressive model method solves the problem,caused by large degrees of freedom of structures,that the matrix of regressive coefficients is not positive definite.The method is efficient in simulating nodal wind speed time series which has temporal and spatial correlation,and that the simulation precision,simulation efficiency and calculation stability can answer the needs of practical engineering.