Uma meta-heurística híbrida com busca por agrupamentos aplicada a problemas de otimização combinatória




- This thesis presents a hybrid method, denominated Clustering Search (CS), that consists of detecting dynamically promising regions in the search space based on the frequence that are sampled in these regions the solutions originated from the metaheuristic. A iterative clustering process is executed in ensembling the metaheuristic, grouping the similar solutions and keeping solutions that are representative to the clusters. The promising regions must be explored as soon as they are discovered, by means of local search heuristics. Some applications of CS are proposed in different combinatorial optimization problems found in literature like the Capacitated p-Median Problem, Capacitated Centred Clustering Problem, Prize Collecting Traveling Salesman Problem and the Assembly Line Worker Assignment and Balancing Problem. These problems have different characteristics and particularities, therefore, it is possible to analyse the behavior of CS in several situations. In these approaches different metaheuristics are utilized to generate solutions for the clustering process of CS, and also a generator method of random solutions. The computational tests present the potential of CS for resolving these optimization problems, putting it as an alternative for the problems that demand to be solved in an approximate form and in a competitive computational time. Conclusions regarding the components and parameters of CS are also presented.


clustering search combinatorial optimization metaheuristics busca por agrupamentos traveling salesman location of facilities assembly line meta-heurísticas localização de facilidades caixeiro viajante linha de produção otimização combinatória

Documentos Relacionados