MAThEMATIcal METhods
The use of expensive equipment does not guarantee the quality of measurements.
A good design and thorough understanding of the measurement, the establishment of a suitable measurement model, and a proper analysis of the data taken are the key to create confidence in measurements.
In order to support our various measurement activities, DFM carries out research on design and modeling of measurements and the optimum analysis of the data taken. A general, non-linear least squares method developed by DFM is now implemented in routine calibration work and is used in various research projects. This method provides the best estimates and the associated covariance matrix of all relevant quantities in a measurement. The method includes an overall test for consistency between data and model and an identification of potential outliers.
As a spin-off of the research, Excel based software tools for the evaluation of measurements have been developed, e.g. DFM-GUM© and DFM-LSQ Analyzer©.
In addition, DFM have worked extensively with validation of metrological software, which has been a popular subject in training courses organized in collaboration with the national accreditation body DANAK and the laboratory union EUROLAB Denmark.