A Mathematical Transform to Analyze Part Surface Quality in Manufacturing

摘要

The status of fault patterns on part surfaces canprovide valuable information about the condition of amanufacturing systems. Accurate detection of the part surfacecondition in manufacturing ensures the fault-free manufacturingof high-quality parts, as well as helping in the accuratedesign/redesign of machine components and manufacturingparameters. To address this problem, we introduce an alternativemathematical transform that has the potential to detect faults inmanufacturing machines by decomposing signals into individualcomponents. Specifically, the paper focuses on thedecomposition of numerically menerated data using theKarhunen-Loeve transform to study a variety of signals frommanufacturing. The potential utility of the proposed technique isthen discussed in the context of understanding a manufacturingprocess under constant development.