Full factorial experimental design for parameters selection of Harmony Search Algorithm

Abstract

Abstract

Metaheuristic may be defined as an iterative search process that intelligently performs the exploration and exploitation
in the solution space aiming to efficiently find near optimal solutions. Various natural intelligences and inspirations have been artificially embedded into the iterative process. In this work, Harmony Search Algorithm (HSA), which is based on the melody fine tuning conducted by musicians for optimising the synchronisation of the music, was adopted to find optimal solutions of nine benchmarking non-linear continuous mathematical models including two-, three- and four-dimensions. Considering the solution space in a specified region, some models contained a global optimum and multi local optima. A series of computational experiments was used to systematically identify the best parameters of HSA and to compare its performance with other metaheuristics including the Shuffled Frog Leaping (SFL) and the Memetic Algorithm (MA) in terms of the mean and variance of
the solutions obtained.