The requirement

Many metrology areas depend on computation. Traceability requires that measurement results can be linked to references through a documented, unbroken chain. In order to ensure traceability, it is necessary that all computational links in the traceability chain are recognised explicitly and known to be operating correctly.

JRP objectives and challenges

To develop new technology that will deliver traceability of computationally-intensive metrology, transparently and efficiently, at the point of use, every time a metrology software component is used:

  • Identify and prioritise key areas of computationally-intensive metrology.
  • Develop a rigorous mechanism for specifying computational aims in clear, unambiguous terms.
  • Develop data generators to produce reference data and results associated with computational aims.
  • Develop criteria to assess the fitness for purpose of computational software in metrology.
  • Develop an enabling ICT (information and communication technologies) infrastructure capable of establishing traceability at the point of use.

The context

ICT has a dominating impact in all aspects of enterprise and quality of life, and many areas of metrology rely heavily on computation. Traceability, measurement standards, and quality systems all demand that computational links are demonstrated to be fit for purpose, but:

  • There is no coherent framework for testing metrology software. Software developers and measuring equipment suppliers have to fill these gaps with their own ad hoc approaches.
  • For software performing complex computation, adequate testing is difficult without an effective method of knowing if the software is producing accurate results.
  • Computational software depends on finite-precision arithmetic, and it is not sufficient to claim that the software is bug-free.
  • For difficult computations, approximate solution methods are used but there is no way of demonstrating that such methods provide results that are fit for purpose.

A new approach

TraCIM will deliver a new approach to validating metrology software, using new mathematics, numerical analysis and exploiting state-of-the-art ICT:

  • Formal and complete statement of computational aims, making it clear what computation task the software is addressing, using a correct mathematical foundation, reducing ‘specification uncertainty’.
  • The self-validating data generator concept. The data generator concept uses the optimality conditions associated with a computational problem to generate data for which the corresponding solution is already known. Customised data sets will be developed for the user’s applications.
  • Performance metrics for metrology software. Fitness for metrology purpose is assessed in terms of the uncertainty contribution of the software to the overall uncertainty of measurement results.
  • Exploiting internet services. Internet-based services will be used to deliver software validation at the point of use.

Delivering impact

  • Calibration in the digital era. TraCIM will put in place, for the first time, a coherent framework for ensuring traceability in computationally-intensive metrology, a basis for ensuring the trustworthiness and fitness for purpose of metrology software for coming decades.
  • A paradigm shift, led by key case study applications. TraCIM will be demonstrated by complete, fully operational systems in case study applications. Partners Hexagon, Mitutoyo, Werth and Zeiss will deliver this new capability to thousands of users, changing the traceability landscape for computation in coordinate metrology and providing a convincing pathway for other metrology domains.
  • Enhancing confidence in metrology products. Users will be able to demonstrate to themselves that the computational software is appropriate for their computational task and their data. Metrology system suppliers will have an independent assessment of the validity of their software.

Work package structure

Project activities to date

  • Twenty metrology applications in four metrology domains have been identified for prioritisation. They include computations associated fitting geometric surfaces to coordinate data according least squares, Chebyshev and related criteria (length), the analysis of spectroscopic data (quantity of matter/chemistry), vector network analysis (electromagnetics), regression, and uncertainty evaluation (interdisciplinary metrology). This last activity is being augmented by a 10 month Early Stage Research Mobility Grant that started in March 2013.
  • A methodology for the formal specification of computational aims has been completed along with a database designed for the capture of such formal specifications. A research excellence grant for further research into the use of formal specification languages for encoding computational aims is anticipated.
  • A scheme for the generation of reference datasets (input data and solution parameters) associated with a range of least squares calculations has been completed. The partners are undertaking research to derive an analogous scheme for Chebyshev approximation; see, A B Forbes and H D Minh, Generation of numerical artefacts for geometric form and tolerance assessment, Int. J. Metrol. Qual. Eng., p145–150, 2012.
  • Project partners, led by PTB, are working hard on the development of the ICT infrastructure that will deliver the TraCIM system.
  • Project scientists Dr Frank Härtig, PTB, and Prof Alistair Forbes, NPL, chaired a joint TC14–TC21 session on Computation in Coordinate Metrology at the IMEKO XX World Congress in Busan, S Korea in September, 2012. The session included two presentations from project partners.