Algoritmo de enxame de partículas para resolução do problema da programação da produção Job-shop flexível multiobjetivo

• Ano de defesa: 2016
• Autor(a) principal: Aranha, Gabriel Diego de Aguiar
• Orientador(a): Kato, Edilson Reis Rodrigues
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação - PPGCC
• Palavras-chave em Português: Enxame de partículas; Arrefecimento simulado; Otimização multiobjetivo; Roteamento; Programação
• Palavras-chave em Inglês: Particle swarm; Simulated annealing; Optimization multi-objective; Routing; Scheduling
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/8861

The companies today are looking for ways to expand their competitive advantages, optimizing their production, and in this context, they found solutions in activities of production scheduling. The production scheduling of the type job-shop, results in one of the most complex problems of combination, the Job-shop Scheduling Problem (JSP), which deterministic resolution is not feasible in polynomial computational time. The Flexible Job-shop Scheduling Problem (FJSP) is a classic extension of the JSP and has been widely reported in the literature. Thus, optimization algorithms have been developed and evaluated in the last decades, in order to provide more efficient production planning, with emphasis to artificial intelligence algorithms of the swarm type, that the latest research presented favorable results. The FJSP allows an operation to be processed for any machine arising from a set of machines along different routes. This problem is commonly dismembered into two sub-problems, the assignment of machines for operations, which is called routing, and operation scheduling. In the FJSP context, this research presents the resolution of the FJSP multi-objective, using a hierarchical approach that divides the problem into two subproblems, being the Particle Swarm Optimization (PSO), responsible for resolving the routing sub-problem, and tasking three local search algorithms, Random Restart Hill Climbing (RRHC), Simulated Annealing (SA) and Tabu Search (TS), for the resolution of scheduling sub-problem. The implementation of the proposed algorithm has new strategies in the population initialization, displacement of particles, stochastic allocation of operations, and management of scenarios partially flexible. Experimental results using technical benchmarks problems are conducted, and proved the effectiveness of the hybridization, and the advantage of RRHC algorithm compared to others in the resolution of the scheduling subproblem.