High-Definition Metrology Based Root Cause Diagnosis Using Root Cause Diagnosis Using 2-D Discrete Cosine Transform

摘要

Surface quality plays an important role in the performance of machined parts. It is therefore critical to inspect the surface variation in details and identify the variation sources. High-definition metrology (HDM) based on laser holographic interferometer makes a breakthrough in surface measurement by quickly scanning an area up to 300×300 mm~2 with micron-level lateral resolution and sub-micron height resolution. Such HDM data usually contain a mixture of various deformation patterns due to a number of variation sources and it becomes challenging to identify each of them for root cause diagnosis. Conventional methods either focus on image decomposition or surface deformation prediction via engineering modeling. However, the former approach faces the challenges of obtaining physically meaningful patterns while the latter shows its limitations in effectively distinguishing patterns, e.g., when global and local deformation patterns both present. This paper proposes a new approach to diagnose variation patterns at different scales as reflected from HDM data based on 2D Discrete Cosine Transform (DCT). The 2D DCT generates orthogonal patterns that are jointly used to estimate surface variation patterns. A statistical diagnostic algorithm is then used to evaluate the significance of the extracted patterns. The proposed approach is demonstrated through a case study using a two-stage machining experiment.