Customization of PSO Algorithm for Fast Design of Emergency Medical System

Abstract

Mathematical modelling, and integer programming generally, has many practical applications in the field of emergency system designing. Effective and fast solving approaches for real-sized problems constitute a key stone of the decision support systems. This paper is focused only on a special class of location problems in which the exact number of facilities are to be selected in order to minimize average response time of the emergency medical system. Since the exact methods are not suitable for their unpredictable computational time, we focus on usage of a special type of a particle swarm optimization algorithm transformed to a discrete mode. We aim at the goal to confirm the hypothesis that it is possible to suggest a sophisticated heuristic for the problem, which can produce a near-to-optimal solution in much smaller time than demanded by exact methods. Especially, we concentrate our effort on customization of the algorithm by usage of uniformly deployed set as a starting swarm and on determining its most convenient size. To verify the impact of the customization, computational study with real-sized benchmarks was performed.