A novel grouping heuristic algorithm for the switch location problem based on a hybrid dual harmony search technique

Abstract

This manuscript proposes a novel iterative approach for the so-called Switch Location Problem (SLP) based on the hybridization of a group-encoded Harmony Search combinatorial heuristic (GHS) with local search and repair methods. Our contribution over other avantgarde techniques lies on the dual application of the GHS operators over both the assignment and the grouping parts of the encoded solutions. Furthermore, the aforementioned local search and repair procedures account for the compliancy of the iteratively refined candidate solutions with respect to the capacity constraints imposed in the SLP problem. Extensive simulation results done for a wide range of network instances verify that statistically our proposed dual algorithm outperforms all existing evolutionary approaches in the literature for the specific SLP problem at hand. Furthermore, it is shown that by properly selecting different yet optimized values for the operational GHS parameters to the two parts comprising the group-encoded solutions, the algorithm can trade statistical stability (i.e. lower standard deviation of the metric) for accuracy (i.e. lower minimum value of the metric) in the set of performed simulations. © 2011 Springer-Verlag.