Optimización de la cadena de suministro mediante el uso de un algoritmo genético basado en clusterización

Abstract

This research proposes k-NSGA-II, an algorithm based on evolutionary computing that mixes properties of micro-algorithms and artificial intelligence fundamentals such as clustering. The objective of k-NSGA-II is to optimize processes in real environments and to efficiently support decision making in organizations. The changes developed were made in NSGA-II, a multiobjective genetic algorithm based on the non-dominance of its results. The functionality of k-NSGA-II was verified by performance tests comparing it with NSGA-II and µ-NSGA-II. These tests were performed on different objective functions and on a case study, which was based on the optimization of product production and distribution. The objectives were to minimize waste and maximize profit through sales. k-NSGA-II was used to optimize the functions generating interesting results for the company. It was also more useful with respect to NSGA-II as it generated a small number of accurate solutions that the analyst could review quickly before making a decision, compared to NSGA-II which works with sets of 200 solutions or µ-NSGA-II which did not present solutions that could be useful to the company. The k-NSGA-II algorithm presents an innovation with respect to NSGA-II as it is a precise micro algorithm whose solutions are very useful for decision making in real problem environments. It improves the evaluation time and avoids the analyst's fatigue because it does not present a great amount of results that many times are not analysed.