Parameter adjustment for genetic algorithm for two-level Hierarchical Covering Location Problem

Abstract

In this paper the two-level Hierarchical Covering Location Problem - HCLP is considered. A new genetic algorithm for that problem is developed, including specific binary encoding with the new crossover and mutation operators that keep the feasibility of individuals. Modification that resolves the problem of frozen bits in genetic code is proposed and tested. Version of fine-grained tournament [5] was used as well as the caching GA technique [12] in order to improve computational performance. Genetic algorithm was tested and its parameters were adjusted on number of test examples and it performed well and proved robust in all cases. Results were verified by CPLEX.