National Physical Laboratory

Traceability for computationally-intensive metrology

Title: Traceability for computationally-intensive metrology
Lead Scientist: Prof Alistair Forbes
Team members: Peter Harris, Keith Lines, Ian Smith
Duration: June 2012 to July 2015
Funding: European Metrology Research Programme / UK National Measurement System Materials and Modelling Programme

Traceability requires that measurement results can be linked to references (such as measurement units) through a documented unbroken chain. If the chain involves computation, as it does in almost all modern measuring systems, it is necessary that all computational links are recognised explicitly and known to be operating correctly. In an era in which metrology was essentially embodied in hardware, establishing traceability was achieved through a series of calibrations, performed according to documentary standards, using reference artefacts. For metrology systems involving significant computation, there are few comparable traceability mechanisms in place.

For measurement results that are determined from significant computation, there can be no real concept of traceability without proper consideration of the computational links in the traceability chain. Developers of metrology software may well work hard to test and validate their software but they have no method of demonstrating to users, through an independent assessment according to internationally accepted procedures, that the software meets the requirements of users. From a user's point of view, they run software supplied as part of measuring systems every day but have no way of knowing if the software is working correctly for their particular application.

Aim of this project is to deliver computational traceability at the point of use. The project is organised into five technical work packages.

  • WP1: Prioritise the areas of metrology to be addressed and consider the commercial and legal issues necessary to deliver impact to the stakeholder community.
  • WP2: Provide formal statements of the computational aims of the software component in clear, unambiguous terms. www.tracim-cadb.npl.co.uk
  • WP3: Generate reference data or 'numerical artefacts' associated with computational aims to test the performance of software.
  • WP4: Develop criteria to assess the performance of software on appropriate numerical artefacts.
  • WP5: Develop an enabling ICT (information and communication technologies) infrastructure to perform the software validation or 'numerical calibration' at the point of use.

The work packages are related as follows:

Traceability for computationally-intensive metrology work packages

A major innovative element of the project is the development of data generators associated with specific computational aims. For most well-defined computational tasks such as fitting a model to data, it is possible to write down optimality conditions, in the form of equations that the solution must satisfy. The data generator concept uses these optimality conditions to generate data for which the corresponding solution is already known. In practice, it is usually much easier to solve the data generation problem than to solve the data fitting problem, which means that reference data can be generated much more efficiently than developing reference software implementations.

The project is a collaboration between NPL (UK), CMI (Czech Republic), INRIM (Italy), PTB (Germany), UM (Slovenia) and VSL (The Netherlands), industrial partners Hexagon, Mitutoyo, Werth and Zeiss. Additional collaborations funded through grants involve the Universities of Huddersfield (UK), Zwickau (Germany), Ostfalia (Germany) and York (UK) and GUM (Poland).

NPL's main research effort in the project relate to:

  • WP2 Computational aim: mathematical analysis.
  • WP3 Data generators: analysis of optimality conditions, analysis numerical stability and sensitivity, development of algorithms for reference data generation.
  • WP4 Performance metrics: numerical stability of algorithms, assessment of numerical uncertainty and simulated measurement uncertainty.

Conference presentations

  • A B, Forbes, Computation and its uncertainty contribution in the traceability chain, AMCTM 2014,
    9-12 September 2014, St Petersburg. (short course).
  • A.B. Forbes, I.M. Smith, F. Härtig, K. Wendt, Overview of EMRP Joint Research Project NEW06 'Traceability for Computationally Intensive Metrology' (TraCIM), AMCTM 2014, 9-12 September 2014, St Petersburg.

Papers and references

  • B. Forbes and H. D. Minh. Generation of numerical artefacts for geometric form and tolerance assessment. Int. J. Metrol. Qual. Eng., pages 145-150, 2012.
  • B. Forbes and H. D. Minh. Form assessment in coordinate metrology. In E. H. Georgoulis, A. Iske and J. Levesley, editors, Approximation Algorithms for Complex Systems, Springer Proceedings in Mathematics, Vol 3, 69-90, Heidelberg, Springer-Verlag (2011)
  • M. G. Cox, M. P. Dainton, A. B. Forbes, and P. M. Harris. Validation of CMM form and tolerance assessment software. In G. N. Peggs, editor, Laser Metrology and Machine Performance V, pages 367-376, Southampton, 2001. WIT Press.
  • M. G. Cox and A. B. Forbes. Strategies for testing form assessment software. Technical Report DITC 211/92, National Physical Laboratory, Teddington, December 1992.
  • B. Forbes, P. M. Harris, and I. M. Smith. Correctness of free form surface fitting software. In D. G. Ford, editor, Laser Metrology and Machine Performance VI, pages 263-272, Southampton, 2003. WIT Press.
Last Updated: 15 Jul 2014
Created: 15 Jul 2014

Registration

Please note that the information will not be divulged to third parties, or used without your permission

Login