Repeatability/Linearity Study and Measurement Loss Cost Analysis of an Automatic Gauge

Computer-aided inspection (CAI) has become one of the fundamental steps in advanced manufacturing by employing automatic gauges. However, it is necessary to provide a more practical way to evaluate CAI on-line efficiency. Since a gauge capability study should be conducted before applying any automatic gauge, the necessary off-line gauge assesment for CAI is introduced in this paper. An R&R (repeatability/reproducibility) study is commonly conducted in the gauge capability study, but in this paper, an R&L (repeatability/linearity) study of a machine vision system is investigated and introduced. The ANOVA approach and regression approach are applied and compared to study the linearity effect. The measurement loss cost includes extra costs (i.e. costs of discarding, reworking, inspection, and quality loss) that arise from measurement error. These costs are analysed by interpreting the relationships between manufacturing capability, gauge capability, and the tolerance. As most gauge capability studies were conducted for the bilateral tolerance inspection, unbalanced tolerance inspection is considered here. The way that the relative costs contribute to the measurement loss cost is also studied. Rather than using a constant measurement error determined by previous observations, the measurement error model is embedded in the relative cost models. Users can then evaluate CAI on-line efficiency by incorporating the measurement error when the tolerances are rapidly changed to satisfy customer requirements.