Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.769071
Title: Combined numerical and statistical modelling for in-depth uncertainty evaluation of comparative coordinate measurement
Author: Papananias, Moschos
ISNI:       0000 0004 7656 6115
Awarding Body: University of Huddersfield
Current Institution: University of Huddersfield
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Quality assurance at low cost needs a tight interaction between machining and inspection. For this reason, the modern view of quality control (QC) requires highly repeatable coordinate measuring systems (CMSs) capable of being integrated into the manufacturing process for in-process feedback. Using this method, it becomes possible to reduce scrap levels and production costs while increasing part throughput. CMSs such as coordinate measuring machines (CMMs) have been used for decades in traditional manufacturing industry to ensure that the size and form of a part conform to design specifications. Although CMMs are considered as powerful and accurate measuring systems, most can only maintain or guarantee their measurement capability in quality control rooms typically having environmental temperature control systems set to maintain a nominal 20°C and maximum variation typically limited to ±2°C.However, shop floor environments have significant variability in ambient temperature. The need in manufacturing for dimensional inspection on the shop floor has led to many technological advancements in manufacturing metrology during recent years. In particular, a recent development includes a parallel kinematic machine (PKM)-based automatic flexible gauge, which is the system under investigation for this thesis. In order to be able to determine the measurement capability of a measuring or gauging machine to dimension a part reliably, it is necessary to evaluate the measurement uncertainties. This thesis first employs the design of experiments (DOE) approach to implement a practical analysis of measurement uncertainty of the automated flexible gauge. Several experimental designs are applied to investigate the influence of various key factors and their interaction on the uncertainty associated with coordinate measurement in comparator mode, in which the geometry of a part is compared with that of a calibrated master part nominally of the same shape. The ISO 15530-3 method is applied to derive uncertainty budgets for the flexible gauging system. A comparison is then made between typical shop floor measurement methods namely hard gauging, on-machine probing (OMP) and the automated flexible gauge. A set of identical test pieces was manufactured and then measured repeatedly using each method, with process and operator variability added as necessary to include typical industrial conditions. The measurement uncertainty is then calculated and compared for each of the measurements. The results show the measurement uncertainty of the comparator technique, which are lower than would be expected from an absolute measurement under workshop conditions. Finally, Markov chain Monte Carlo (MCMC) methods are applied to evaluate uncertainty associated with comparative coordinate measurement using a more realistic probability model to avoid repeating measurements. Samples are drawn from the unnormalized posterior using Gibbs sampling. Another feature of this thesis is the developed empirical method based on Bayesian regularized artificial neural networks (BRANNs) for estimating point coordinates and associated uncertainties when no satisfactory measurement model can be developed and large experimental designs are not practical. The effectiveness of the proposed method is demonstrated using two case studies.
Supervisor: Fletcher, Simon ; Longstaff, Andrew Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.769071  DOI: Not available
Keywords: T Technology (General) ; TA Engineering (General). Civil engineering (General)
Share: