A Modification of Artificial Bee Colony Algorithm Applied to Loudspeaker Design Problem

Abstract

The design of electromagnetic devices is generally multimodal, multidimensional, and constrained. Its simulation-based feature makes traditional optimization algorithms unsuitable to use; whereas metaheuristic approaches are good choices for tackling such problems. Since many electromagnetic design problems require excessive computer resources even on modern computers, attempts to improve the efficiency and effectiveness of metaheuristic approaches are very challenging for designers. This paper proposes a modification of artificial bee colony (ABC) algorithm. The design philosophy of this algorithm is to promote the convergence rate of ABC. The properties of the modified ABC algorithm are analyzed by testing, first, on two mathematical functions. The proposed algorithm is then applied to solve a loudspeaker design problem using FEM. Promising performance is obtained when compared with those using original ABC, genetic algorithm, particle swarm optimization, and differential evolution algorithms.