Published: 2008 März
Institution: University of Karlsruhe, Institute AIFB
Erscheinungsort / Ort: 76128 Karlsruhe, Germany
Many practical optimization problems change over time, requiring a repeated re-adaptation of the solution. As has been shown in numerous papers, evolutionary algorithms (EAs) can be modified so that they can cope well with dynamic environments. However, basically all papers so far either considered a continuous change, or a change that happens in intervals which coincide with the time to complete a number of generations. In reality, changes may occur at any time and irregular intervals, which raises the question how to deal with a change occurring within a generation of an EA. Different methods for handling such changes are given and experimentally compared in this study.