TIME SERIES ANALYSIS OF STANDARD SPHERE CMM DATA (SPHERICITY) FOR MACHINE LEARNING

Abstract

The coordinate measuring machines (CMM’s) has given a new impulse in the field of geometrical and dimensional metrology.The CMM’s in industrial environments have become an important resource for the quality systems, monitoring manufacturingprocesses, reduction errors during the manufacturing process, inspection of product specifications and in continuous qualityimprovement. However, there is a need to evaluate, through practical, fast, effective and low cost methods, the CMMmetrologicalspecifications.In this research work, time seriesanalysis of standard sphere CMM data (Sphericity) for machine learning purpose is carried out by a comparative study between the precision(repeatability)obtained withsphericity measurement of three different(in diameter)probe calibration spheresby three different(in diameter)combination ofstyli, measuredona specific coordinate measuring machine (CMM). Styliused for this study areof ruby material and sizesare 2X30 mm; 3X30 mm & 4X30 mm (Tip Diameter X Shank length mm). Probe calibration spheres used for this study are of ceramic material and diameter sizes are 19.9705mm;19.9940mm&29.9971mm.These threetypes of probesand calibration sphereare most commonly used in CMMs. In order to thistime series analysis,three standard ceramic spheres of diameter 19.9705 mm, 19.9940mm and 29.9971 mmweremeasuredfor sphericityat 9 distinct locations in the CMM working volume with three different stylusesof 2X30 mm; 3X30 mm & 4X30 mm.The above-mentionedsphericitymeasurements were repeated three times at each location and exercise was conducted for 10 days.All themeasurements were taken at a temperature of 20 ºC ± 1 ºC.