Accurate models of technical systems are the basis for many tasks like system analysis, predictions, or controller design. Usually, the values of several important parameters cannot be determined by theoretical analysis only; instead, process identification is necessary. For several applications, the efficiency of the identification procedure is very important, for example, for the creation of thermal models of machine tools, because of the large time constants and the expensive machine time. The goal of the authors is the support of this task as far as possible by software. This paper contributes to that goal twofold: on the one hand, it provides a collection of influences which have to be considered for supporting the identification procedure. On the other hand, concepts for computer-based support are presented-ontologies and automatic design methods based on evolutionary algorithms.