Design of general-purpose sampling strategies for geometric shape measurement

Abstract

Quality inspection is a preliminary step for different further analyses (process monitoring, control and optimisation) and requires one to select a measuring strategy, i.e., number and location of measurement points. This phase of data gathering usually impacts on inspection times and costs (via sample size) but it also affects the performance of the following tasks (process monitoring, control and optimisation). While most of the approaches for sampling design are specifically presented with reference to a target application (namely, monitoring, control or optimisation), this paper presents a general-purpose procedure, where the number and location of measurement points are selected in order to retain most of the information related to the feature under study. The procedure is based on principal component analysis and its application is shown with reference to a real case study concerning the left front window of a car. A different approach based on multidimensional scaling is further applied as validation tool, in order to show the effectiveness of the PCA solution.