Deterministic parameter control in Harmony Search

Abstract

Harmony search is a recently developed meta heuristic capable of solving discrete and continuous valued optimisation problems. However, the nature of pre-defined constant parameters limits the exploitation of the algorithm. This paper proposes a number of deterministic parameter control rules to fine-tune these parameters individually and dynamically, making Harmony Search a more dynamic algorithm which is able to achieve better results. A combined approach that implements all the proposed rules is then applied to various benchmarks and engineering problems. Experimental results reveal that the combined approach can find better solutions when compared to the original harmony search and several other heuristics, making harmony search a strong mechanism to perform optimisation tasks.