Errors produced in simulating a stationary Gaussian multi -variate wind process, with the original spectral representation method (OSRM) and the proper orthogonal decomposition - based one (PSRM) were assessed. To begin with, a set of formulas describing the two types of SRM and the closed - form stochastic errors produced with them were presented in a standard form. Accordingly, the errors produced in simulating a 64 - point simplified wind field using PSRM and OSRM respectively were computed and then compared in terms of the global spectral stochastic errors and the relative ones of standard deviations. It was concluded from the numerical example that the errors produced with PSRM are not only less in total but also distributed more uniformly to every point than those with OSRM; thus PSRM should be preferred to. In addition, the influence of the point number of frequency sampling on the errors was discussed in detail.Some approaches for relieving the errors were also suggested.%对原型和POD型谱表示法模拟多变量正态风场时的随机误差进行了对比分析.给出了原型,POD型谱表示法的风场模拟标准算法,以及使用这套标准算法模拟风场时所产生随机误差的解析计算公式.基于此解析解,以总体随机误差和根方差相对随机误差为标准,比较了两类谱表示法在模拟一个64点简单风场时的随机误差.数值分析结果表明:相对于原型谱表示法,POD型谱表示法的模拟随机误差较小而且空间分布更加均匀,模拟时应优先选用.此外,还研究了频率采样点数对误差的影响;给出了控制模拟随机误差的方法.
展开▼