ENHANCING THE QUALITY AND RELIABILITY OF INSPECTION DATA FOR PRECISION MANUFACTURING

摘要

Dimensional inspections are commonly used to scrutinize the quality of manufactured products against established standards and specifications. However, many inspection processes are contaminated by various measurement errors. One of the prominent sources for measurement errors is due to the imperfection of a measuring device and the compound effect of its imperfection with geometric characteristics of a measured feature. To ensure the quality and reliability of any inspection process, measurement errors need to be identified and reduced by minimizing the effect of the compound errors. If this can be done, the quality of collected data can be enhanced and a more meaningful analysis result can then be drawn. There is also an urgent need to derive sampling strategy for specifying a set of measuring points that lead to accurate sampling while minimizing sampling time and cost. Issues of measurement error identification and reduction are discussed. Analytical models are derived to assess and decouple the compound effect of errors, and thus, reduce the measurement errors. Then, feature-based Hammersley sequence and stratified sampling methods are used to derive effective sampling strategy for various geometric features.