High-resolution spatial data are essential for characterizing and monitoring surface quality in manufacturing. However, the measurement of high-resolution spatial data is generally expensive and time-consuming. Interpolation based on spatial models is a typical approach to cost-effectively acquire high-resolution data. Conventional modeling methods fail to adequately model the spatial correlation induced by periodicity, and thus their interpolation precision is limited. In this paper, we propose using a Bessel additive periodic variogram model to capture such spatial correlation. When combined with kriging, a geostatistical interpolation method, accurate interpolation performance can be achieved for common periodic surfaces. In addition, parameters of the proposed model provide valuable insights for the characterization and monitoring of spatial processes in manufacturing. Both simulated and real-world case studies are presented to demonstrate the effectiveness of the proposed method.
Read the paper here.