A comparative performance analysis of intelligence-based algorithms for optimizing competitive facility location problems

Abstract

Most companies operate to maximize profits and increase their market shares in competitive environments. Since the proper location of the facilities conditions their market shares and profits, the competitive facility location problem (CFLP) has been extensively applied in the literature. This problem generally falls within the class of NP-hard problems, which are difficult to solve. Therefore, choosing a proper solution method to optimize the problem is a key factor. Even though CFLPs have been consistently solved and investigated, an important question that keeps being neglected is how to choose an appropriate solution technique. Since there are no specific criteria for choosing a solution method, the reasons behind the selection approach are mostly unclear. These models are generally solved using several optimization techniques. As harder-to-solve problems are usually solved using meta-heuristics, we apply different meta-heuristic techniques to optimize a new version of the CFLP that incorporates reliability and congestion. We divide the algorithms into four categories based on the nature of the meta-heuristics: evolution-based, swarm intelligence-based, physics-based, and human-based. GAMS software is also applied to solve smaller-size CFLPs. The genetic algorithm and differential evolution of the first category, particle swarm optimization and artificial bee colony optimization of the second, Tabu search and harmony search of the third, and simulated annealing and vibration damping optimization of the fourth are applied to solve our CFLP model. Statistical analyses are implemented to evaluate and compare their relative performances. The results show the algorithms of the first and third categories perform better than the others.