Importance of Parameterization to Improve Meta-heuristics Performance for Smart Grid Applications

Abstract

The development of the Smart Grid requires handling big amounts of data and solving large combinatory problems under uncertainty. Meta-heuristic algorithms are a useful tool to handle such problems. One of the main problems of these algorithms is that they have to operate in real time. Thus, it is necessary to reduce computation times as much as possible. This paper demonstrates that an extensive empirical study of parameterization leads to a very significant improvement of algorithm efficiency. It focuses on Item Oriented Ant Colony Optimization (IOACO) algorithm reducing by one third the computation time of the most efficient method in the references.