Browsing by Author "Sotamba Once, Luis Miguel"

Now showing 1 - 2 of 2

Desarrollo de un algoritmo híbrido metaheurístico para el problema de distribución de planta en las MIPYMES textiles del Ecuador
(Universidad de Cuenca, 2023-09-20) Sotamba Once, Luis Miguel; Sigüenza Guzmán, Lorena Catalina
The Facility Layout Problem (FLP) refers to finding the most effective arrangement of facilities in a factory, considering various aspects. These facilities could include departments, personnel, or machinery, to name a few examples. The FLP is an NP-Hard problem, meaning that there are no algorithms capable of providing an optimal solution in a reasonable polynomial time. Due to this, researchers have turned to metaheuristics and have attempted to combine them to create hybrids. Hybrid metaheuristics can leverage the strengths of each approach to generate higher-quality solutions that are close to optimal. However, many studies that develop hybrid metaheuristics do not use a methodology that explains the development process. For this reason, this work focuses on composing a hybrid metaheuristic following a methodology that encourages the use of SWOT analysis, the Strategic Choice Approach, and Convergent/Divergent Thinking. The hybrid metaheuristic called GENTSA was constructed using a Genetic Algorithm, Simulated Annealing, and Tabu Search. GENTSA solves the FLP that arises in small and medium-sized textile businesses in Ecuador, which was used as a case study. Additionally, as part of its validation, test functions obtained from the state of the art were used. GENTSA was compared to other metaheuristics where it achieves the best facility layout with the lowest cost and reasonably adapts to problem domains beyond its original design scope.
Driver Analysis to Solve Dynamic Facility Layout Problems: A Literature Review
(Springer, 2023) Sotamba Once, Luis Miguel; Peña Ortega, Mario Patricio; Sigüenza Guzmán, Lorena Catalina
The Dynamic Facility Layout Problem (DFLP) is concerned with finding an optimal facility design considering changes in the planning horizon. Since DFLP belongs to the non-polynomial hard class problem, different solutions have been used to find an optimal solution. However, a correct performance evaluation is needed to validate and compare the results with others. This performance evaluation refers to using both statistical and computational tests. When searching the literature on related papers, none consider these tests. The lack of such information and the constant evolution of algorithms motivated this work. The current document reviews the solution methodologies applied to solve DFLP and the manner the performance evaluation is done. In addition, the methods used to mix solution methodologies, called hybrid approaches, are included. This work was carried out using the Barbara Kitchenham methodology, in which studies from 2015 to 2022 were considered. A sample of 59 articles was analyzed, all about DFLP. As a result, this study identified two commonly used categories when solving DFLPs: hybrid and metaheuristic approaches. Furthermore, performance evaluation is done using different statistical methods in some cases, comparisons of some numerical results obtained from the algorithm output, and studies without comparisons. Finally, the results do not find any instances in which a methodology is applied to compose the algorithm when a hybrid approach is used. To the best of our knowledge, this work is the first in which performance evaluation is considered.