Evaluation of shape measurements by using non-orthogonal sine functions

Abstract Problems and experience with the development of an improved evaluation algorithm for shape measurements are discussed. The model bases are deterministic as well as stochastic shares as parts of the measuring value sequences. The application of a new approximation method with non-orthogonal sine functions is shown for different evaluation tasks. The new method avoids in principle a series of disadvantages of the usual evaluation methods connected with the finite measuring length and the finite number of measuring values. Taking two examples from measurement practice, the experience gained is shown. The first example results in a substantial data reduction and a significant reduction of measuring uncertainty. The second example is an extensive analysis of a surface profile. It becomes apparent that the typical periodic profile of a milled surface is superimposed by a considerable stochastic share.