Yazar "Sonmez, Rifat" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • Küçük Resim Yok
    Öğe
    Hybrid Genetic Algorithm with Simulated Annealing for Resource-Constrained Project Scheduling
    (Asce-Amer Soc Civil Engineers, 2015) Bettemir, Onder Halis; Sonmez, Rifat
    Resource-constrained project scheduling problem (RCPSP) is a very important optimization problem in construction project management. Despite the importance of the RCPSP in project scheduling and management, commercial project management software provides very limited capabilities for the RCPSP. In this paper, a hybrid strategy based on genetic algorithms, and simulated annealing is presented for the RCPSP. The strategy aims to integrate parallel search ability of genetic algorithms with fine tuning capabilities of the simulated annealing technique to achieve an efficient algorithm for the RCPSP. The proposed strategy was tested using benchmark test problems and best solutions of the state-of-the-art algorithms. A sole genetic algorithm, and seven heuristics of project management software were also included in the computational experiments. Computational results show that the proposed hybrid strategy improves convergence of sole genetic algorithm and provides a competitive alternative for the RCPSP. The computational experiments also reveal the limitations of the project management software for resource-constrained project scheduling. (C) 2014 American Society of Civil Engineers.
  • Küçük Resim Yok
    Öğe
    A Memetic Algorithm for the Solution of the Resource Leveling Problem
    (Mdpi, 2023) Iranagh, Mehdi; Sonmez, Rifat; Atan, Tankut; Uysal, Furkan; Bettemir, Onder Halis
    In this paper, we present a novel memetic algorithm (MA) for the solution of the resource leveling problem (RLP). The evolutionary framework of the MA is based on integration of a genetic algorithm and simulated annealing methods along with a resource leveling heuristic. The main objective of the proposed algorithm is to integrate complementary strengths of different optimization methods and incorporate the individual learning as a separate process for achieving a successful optimization method for the RLP. The performance of the MA is compared with the state-of-the-art leveling methods. For small instances up to 30 activities, mixed-integer linear models are presented for two leveling metrics to provide a basis for performance evaluation. The computational results indicate that the new integrated framework of the MA outperforms the state-of-the-art leveling heuristics and meta-heuristics and provides a successful method for the RLP. The limitations of popular commercial project management software are also illustrated along with the improvements achieved by the MA to reveal potential contributions of the proposed integrated framework in practice.