Regression analysis of automated measurment systems

Abstract

Automated measurement systems are dependent upon successfully application of multiple integrated systems to perform measurement analysis on various units-under-tests (UUT)s. Proper testing, fault isolation and detection of a UUT are contingent upon accurate measurements of the automated measurement system. This paper extends previous presentation from 2007 AUTOTESTCON on the applicability of measurement system analysis for automated measurement systems. The motivation for this research was to reduce risk of transportability issues from legacy measurement systems to emerging systems. Improving regression testing utilizing parametric metadata for large scale automated measurement systems over existing regression testing techniques which provides engineers, developers and management increased confidence that mission performance is not compromised. The utilization of existing software statistical tools such as Minitab <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sup> provides the necessary statistical techniques to evaluate measurement capability of automated measurement systems. By applying measurement system analysis to assess the measurement variability between the US Navypsilas two prime automated test systems the Consolidated Automated Support System (CASS) and the Reconfigurable-Transportable Consolidated Automated Support System (RTCASS). Measurement system analysis shall include capability analysis between one selected CASS and RTCASS instrument to validate measurement process capability; general linear model to assess variability between stations, multivariate analysis to analyze measurement variability of UUTs between measurement systems, and gage repeatability and reproducibility analysis to isolate sources of variability at the UUT testing level.