| || |
Huber, M.; Baier, H.:
In: Proc. of CAD-FEM Users’ Meeting, Bonn, Deutschland, 2005
It is a challenging task in multidisciplinary design optimization to take all important influences on a structure into account. In this paper, we present a method to utilize qualitative human knowledge for disciplines/restrictions for which no numerical simulation code or approximation models from test data exist. Fuzzy rule based systems are used to parameterize the knowledge and experience of domain experts. The approach is applied to a micro car space-frame approximating the according manufacturing restrictions. They are mainly related to the extrusion moulding process of reinforced aluminium and magnesium hollow profiles. Optimal designs with fuzzy geometric restrictions suffer a double increase in mass if the same torsional stiffness of the vehicle structure is demanded. By the additional incorporation of fuzzy manufacturing effort, valuable information for design decisions is generated (e.g. material selection, profile cross section). The advantages of fuzzy rule based systems are the similarity to natural human language for building the mathematical models and the possibility to map also nonlinear functions. The approach is easy to understand for the domain experts. The main challenge of the presented method is the condensation of complex parameter interactions in the considered disciplines.