Multivariate Data Analytics in Surface Topography Assessments: Case Study High Precision Fine Grinding Processes

摘要

The manufacturing of cutlery is a challenging task, where in addition to the functional characteristics the sharpness, the durability, as well as the roughness and the optical perception of the blades play an important role. For this reason, a special high precision fine grinding process has been developed, which is characterised by a large number of process parameters that strongly influence the result of the fine grinded surface. In this study, the influences of product and process parameter sets of the surface topography, which is characterised by the target variables roughness (Raand Rz) respectively the gloss value, are analysed. A selection of influencing product and manufacturing process parameters such as cutlery type and geometry, grinding disc, feed rate and cutting fluid are taken into account. The base of operation are prototype samples, which are manufactured with different process parameter sets. Based on small sample sizes, the distribution model regarding the measurement data of the target variables is unverifiable. Therefore, the data analysis is based on non-parametric statistics, like significance test (random, focus, dispersion, slope and trend). Furthermore, multivariate approaches for the analysis are applied to assess the prediction in relation to the target area specification. This paper concludes with an application of the non-parametric statistics and the multivariate process analysis regarding different cutlery prototypes. The goal is the evaluation of controllability and reproducibility of aperiodic surface topographies of a technically highly challenging fine grinding process.