Model validation and selection based on inverse fuzzy arithmetic

Abstract

In this work, a method for the validation of models in general, and the selection of the most appropriate model in particular, is presented. As an industrially relevant example, a Finite Element (FE) model of a brake pad is investigated and identified with particular respect to uncertainties. The identification is based on inverse fuzzy arithmetic and consists of two stages. In the first stage, the eigenfrequencies of the brake pad are considered, and for three different material models, a set of fuzzy-valued parameters is identified on the basis of measurement values. Based on these identified parameters and a resimulation of the system with these parameters, a model validation is performed which takes into account both the model uncertainties and the output uncertainties. In the second stage, the most appropriate material model is used in the FE model for the computation of frequency response functions between excitation point and three measurement points. Again, the parameters of the model are identified on the basis of three corresponding measurement signals and a resimulation is conducted.