Pictures of the public defense
Members of the exam committee
List of my publications
When trying to model the human perception of environmental noise annoyance, several problems arise that are related to fuzzy, uncertain or lacking data and vaguely known relations between variables. Therefore in this work, the concept of noise annoyance has been considered as an inherent fuzzy concept from the very beginning. This fuzziness and uncertainty is modeled with fuzzy set theory (and related theories), which allows to deal with this kind of information in a very natural and straightforward way. The framework allows to take the inherent fuzziness into account in the representation of noise annoyance as well as in the reasoning processes and the interpretation of the model results.
First, a conceptual model for noise annoyance from a particular source (e.g. road and railway noise) has been studied and implemented as a fuzzy rule based system. This allows experts in the field to formulate their knowledge in almost natural language statements. No knowledge extraction algorithms have been used because the aim is to build a stable model not biased towards a specific data set. Secondly, the accumulation of perceptions of noise annoyance from several sources into a single annoyance expression has been studied. The best known strongest component model is a black box model and therefore theoretically very weak. A cognitive process for this model has been formulated as a set of rules which have also been fuzzified. The accumulation problem has also been approached from the viewpoint of multi-criteria decision making. In this context several non-linear aggregation operators, such as fuzzy integrals, have been adopted to model the accumulation of perceptions.
All models have been tested with data from two social surveys, one conducted in Austria and one from Flanders, Belgium. Results show a better prediction performance than the classical, binary logic based, models. A clear advantage is also their interpretability and their tolerance for fuzzy and uncertain data and knowledge.
Andy Verkeyn's Website, © Andy Verkeyn 1997-2015
Last webpage revision: