Proposal of a hybrid group decision-making model based on dual hesitant fuzzy theory and genetic algorithm

The decision making processes involve several decision makers who evaluate several alternatives in relation to various criteria, which can bring high complexity to the problem. Several studies present multi-criteria techniques based on fuzzy representations for consensus or aggregation of individual judgments. A limitation of these studies is the lack of the combination of consensus approaches and aggregation of individual judgments. The combination of these two types of approaches can contribute to a consensual evaluation process, even if there are divergences due to the different perceptions of managers from different sectors. Thus, the overall objective of this PhD project is to propose a group decision-making model that seeks to combine the consensus and judgment aggregation approaches using the dual hesitant fuzzy sets (DHFS) representation. This representation was selected to overcome the limitations of the intuitionistic fuzzy and hesitant fuzzy representations. The development of the Ph.D. project includes the following phases: literature review; selection of techniques and model construction; computational implementation of the techniques; and pilot application with the evaluation of the proposal. To achieve the stated goal, a decision making model composed of three phases was proposed. The first phase consists in structuring the problem by applying a problem structuring method (PSM). The second phase consists of the proposition of a Genetic Algorithm (GA) to reach consensus among decision makers in the weighting of criteria and decision makers. The third phase consists of the use of DHFS aggregation operators and the PROMETHEE V multicriteria technique to select the portfolio of alternatives. A pilot application was developed in a steel and mining industry to exemplify the use of the proposed model in the context of evaluation and selection of a sustainable actions. Computational tests were performed to evaluate the robustness and performance of the proposed GA by varying the number of criteria, the number of decision makers, the number of hesitant terms, and the required level of consensus. An instance generator was proposed to create the test scenarios. A Particle Swarm Optimization algorithm (PSO) was implemented to allow comparison with the proposed GA. As a result, this study presented a decision-making model that meets the current needs of decision-making processes that 14 involve both the search for consensus and the aggregation of divergent opinions of decision makers. Computational testing and comparison proved the robustness and effectiveness of the proposed consensus GA. The pilot application demonstrated the applicability of the proposed model. As a solution, the portfolio of sustainable actions defined sought to cover the different strategies for improving the sustainability of a supply chain. (AU)