Quantcast

Multi-stage evolution of single- and multi-objective MCLP

Research paper by Helge Spieker, Alexander Hagg, Adam Gaier, Stefanie Meilinger, Alexander Asteroth

Indexed on: 03 Oct '16Published on: 03 Oct '16Published in: Soft Computing



Abstract

Maximal covering location problems have efficiently been solved using evolutionary computation. The multi-stage placement of charging stations for electric cars is an instance of this problem which is addressed in this study. It is particularly challenging, because a final solution is constructed in multiple steps, stations cannot be relocated easily and intermediate solutions should be optimal with respect to certain objectives. This paper is an extended version of work published in Spieker et al. (Innovations in intelligent systems and applications (INISTA), 2015 international symposium on. IEEE, pp 1–7, 2015). In this work, it was shown that through problem decomposition, an incremental genetic algorithm benefits from having multiple intermediate stages. On the other hand, a decremental strategy does not profit from reduced computational complexity. We extend our previous work by including multi-objective optimization of multi-stage charging station placement, allowing us to not only optimize toward (weighted) demand location coverage, but also to include a second objective, taking into account traffic density. It is shown that the reachable part of the full Pareto front at each stage is bound by the solution that was chosen from the respective previous front. By careful choice of the selection strategy, a particular focus can be set. This can be exploited to comply with concrete implementation goals and to adjust the evolved strategy to both static and dynamic changes in requirements.