This page is for old projects and those that are unrelated to my main area of research.

Old Projects

Masters thesis

Written during my studies of Applied Systems Science at the University of Osnabrueck, Germany.

Summary

This work compares neural networks and various mathematical models with respect to their ability to analyze and predict environmental time series data. Applications for environmental modelling include risk assessment for hazardous chemicals, prognosis of concentrations of chemicals in air, water or soil, and others. Most approaches employ analytical mathematical models. Unfortunately, the various variables and their interrelations usually render all but the most complex analytical models useless for environmental modelling. An alternative approach are neural networks, which are specifically designed for situations where the complex interrelations are unknown. The results show that neural networks are a viable alternative to mathematical models under certain conditions. The research itself consisted in analyzing and projecting an environmental time series based on other related series.

Publications

• Hannemann, J. Vergleich Mathematischer Modelle und Neuronaler Netze bei der Analyse and Prognose von Umweltdaten. Diplomarbeit /Masters thesis; Osnabrücker Schriften zur Mathematik, Department of Mathematics and Computer Science, University of Osnabrück, Germany, 1998

 Last Modified: 2007/10/16;