Most modern digital approaches to engineering are based on models and their model transformations. Most of these model transformations are mathematically speaking non-bijective mappings — so-called projections — where some information of the original model is lost during the mapping. From a theoretical point of view it is therefore of great interest to exactly examine the properties of these model transformations. In this paper at first the characteristics of a model are briefly explained. Then some of the most common model-based engineering approaches are reviewed and compared regarding their models and model transformations. In this examination the missing existence of an inverse transformation (a so-called text-to-model transformation, T2M) of a typical model transformation (a so-called model-to-text transformation, M2T) is identified. That discovery may well hold the key to the realization of a so-called round-trip engineering. The required existence of the inverse transformation to this round-trip engineering is then generically postulated as having the nature of a pattern recognition problem. For illustration purposes and a better understanding of the interpretation of the inverse transformation as a pattern recognition problem, a case study for the reconstruction of an abstract model from the concrete model is given using CAD-Data of a satellite. Since CAD models belong to geometry, dimensionless geometric moment invariants play a key role in the generic solution of the pattern recognition problem contained in this example.

This content is only available via PDF.
You do not currently have access to this content.