Comparison of contact conditions obtained by direct simulation with statistical analysis - Normally distributed isotropic surfaces -
Comparison of contact conditions obtained by direct simulation with statistical analysis - Normally distributed isotropic surfaces -
复制标题
直接模拟与统计分析获得的接触条件的比较 - 正态分布各向同性表面 -
DOI：
10.1088/2051-672x/aac072
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Michimasa Uchidate
中科院分区：
文献类型：
--
作者：
多田一生，吉冨健一郎，餅田 正秋，宇根 篤暢;Michimasa Uchidate
在本研究中，为了建立关于粗糙接触分析中山顶提取方法和随机模型选择的影响的系统知识，将不同山顶提取方法的统计接触模型获得的接触面积比（A r/A a）与使用边界元法（BEM）的直接模拟进行了比较。使用在功效指数和相关长度方面具有不同自相关函数的 50 个区域地形数据集进行研究。本研究采用非因果二维自回归模型，可以生成具有指定参数的数据集。检验了三种顶峰提取方法：Nayak 理论、8 点分析和分水岭分割。对于随机模型，应用Bhushan模型和BGT(Bush-Gibson-Thomas)模型。随机模型的 A r/A a 值往往小于 BEM。 Bhushan 的8 点分析模型与BEM 之间的差异略小于Nayak 的理论。分水岭分割的结果与 8 点分析的结果相似。 Wolf 剪枝对随机分析和 BEM 之间差异的影响还不是很清楚。对于采用表面梯度的 BGT 模型，当 Nayak 的带宽参数较大时，可以获得与 BEM 良好的定量一致性。
In this study, with the aim of establishing a systematic knowledge on the impact of summit extraction methods and stochastic model selection in rough contact analysis, the contact area ratio (A r/A a) obtained by statistical contact models with different summit extraction methods was compared with a direct simulation using the boundary element method (BEM). Fifty areal topography datasets with different autocorrelation functions in terms of the power index and correlation length were used for investigation. The non-causal 2D auto-regressive model which can generate datasets with specified parameters was employed in this research. Three summit extraction methods, Nayak's theory, 8-point analysis and watershed segmentation, were examined. With regard to the stochastic model, Bhushan's model and BGT (Bush-Gibson-Thomas) model were applied. The values of A r/A a from the stochastic models tended to be smaller than BEM. The discrepancy between the Bhushan's model with the 8-point analysis and BEM was slightly smaller than Nayak's theory. The results with the watershed segmentation was similar to those with the 8-point analysis. The impact of the Wolf pruning on the discrepancy between the stochastic analysis and BEM was not very clear. In case of the BGT model which employs surface gradients, good quantitative agreement against BEM was obtained when the Nayak's bandwidth parameter was large.