Examinando por Materia "COMBINATORIAL OPTIMIZATION"

Mostrando 1 - 3 de 3
  • Miniatura
    PublicaciónAcceso abierto
    Un algoritmo genético híbrido y un enfriamiento simulado para solucionar el problema de programación de pedidos Job Shop
    (2013-12-17) Meisel-Donoso, José David; Prado, L. K. (Liliana Katherine)
    Job Shop Scheduling Problem (JSP), classified as NP-Hard, has been a challenge for the scientific community because achieving an optimal solution to this problem is complicated as it grows in number of machines and jobs. Numerous techniques, including metaheuristics, have been used for its solution; however, the efficiency of the techniques, in terms of computational time, has not been very satisfactory. Because of this and for contributing to the solution of this problem, a simulated annealing (SA) and an improved genetic algorithm (IGA) have been proposed. The latter, by implementing a strategy of simulated annealing in the mutation phase, allows the algorithm to enhance and diversify the solutions at the same time, in order not to converge prematurely to a local optimum. The results showed that the proposed algorithms yield good results with deviations around the best values found not exceeding 5 % for more complex problems.
  • Miniatura
    PublicaciónAcceso abierto
    Planeamiento multiobjetivo de sistemas de distribución usando un algoritmo evolutivo NSGA-II
    (2013-11-19) López, L. (Libardo); Hincapié, R. A. (Ricardo Alberto); Gallego, Ramón Alfonso
    This paper presents a methodology to solve the problem of distribution system planning, using a multiobjective optimization technique. In the proposed model the location and design of new elements and the upgrading of existing elements are considered. In the problem approach, investment and operation costs and network reliability are included in the objective function. This research topic is relevant to the electricity distribution companies, as it allows for new tools that bring the problem to real situations, such as considering multiple objectives, which has a positive impact on the finances and on operation of the system. In the model solution, an elitist non-dominated sorting algorithm (NSGA-II) is used, and to verify its efficiency a case of specialized literature is used, which corresponds to a distribution network of an electrical system.
  • Miniatura
    PublicaciónAcceso abierto
    Resolución del problema de enrutamiento de vehículos con limitaciones de capacidad utilizando un procedimiento metaheurístico de dos fases
    (2013-11-26) Daza-Escorcia, Julio Mario; Montoya-Torres, J. R. (Jairo Rafael); Narducci-Marín, F. (Francesco)
    This paper presents an alternative procedure to solve the Capacitated Vehicle Routing Problem (CVRP) with homogeneous fleet. The paper proposes a two-phase metaheuristic algorithm: routes design and fleet scheduling. The first phase is based on heuristics and metaheuristics procedures in order to build an initial solution that is then improved using tabu search to obtain non-dominated solutions in polynomial computational time. For the second phase, corresponding to fleet scheduling, the problem is approached using an analogy with the identical parallel machine scheduling problem. This procedure looks for the minimization of the fixed cost of using installed capacity as the objective function. The proposed procedure was tested using both a random-generated instance and real data, giving competitive results in comparison with other heuristics tested.