基于NCGM的粗糙表面数值模拟与实验对比

基于时间序列模型、线性变换方法和Johnson转换系统，研究粗糙表面的重构方法。为减少数据储存空间的需求和收敛性要求，采用非线性共轭梯度法（NCGM）求解自相关矩阵，得到高斯型表面及非高斯型表面的数值模拟方法。对不同相关长度及高度分布的高斯型、非高斯型粗糙表面进行数值模拟，并与实际测量结果进行对比。结果表明，所提出的基于NCGM的粗糙表面数值模拟方法能有效地模拟满足高度分布函数与自相关函数的实际加工表面。

Numerical and Experimental Analysis of Rough Surface Based on NCGM

This paper used the time series model, linear transformation method and Johnson transform system to generate rough surface. In order to reduce the memory requirements and for easy convergence, the NCGM was used to solve the  autocorrelation matrix. To illustrate the numerical procedure, some rough surfaces having Gaussian and nonGaussian height distribution and an exponential autocorrelation function(ACF) were generated. The results show that the method can generate the prescribed height distribution and autocorrelation function based on actual processing the surface effectively.