Including Expert Knowledge in Finite Element Models by Means of Fuzzy Based Parameter Estimation

O. Krol[1], N. Weiss[1], F. Sawo[1], and T. Bernard[1]

[1]Fraunhofer Institute for Information and Data Processing, Karlsruhe, Germany

In this paper we present a novel approach for modeling spatial distributed bio- chemical and environmental processes like the growth of plants and the related biochemical reactions. The physical phenomena like flow and mass transport can be described by fluid dynamics equations, but for effects like growth rates often no analytic models are available. However, in many cases experts have knowledge about the system behavior that can be formulated by a set of if-then-rules. As this kind of knowledge can easily be handled by so-called Fuzzy models we propose the coupling of FEM (Finite Element Method) models with such Fuzzy models. Besides the natural inclusion of expert knowledge a second benefit of this approach consists in the fact that Fuzzy Models can describe even very nonlinear phenomena.

Download