Modelos e métodos de tomada de decisões para apoio à gestão estratégica em empresas de energia elétrica

Abstract

This work presents a wide study on the problems faced by the companies that adopt the Balanced ScoreCard methodology in its activities of strategic management, especially in the decision making processes. As form of minimizing the impacts of the introduction of this methodology, that increases the universe of data and information to be analyzed and still introduces uncertainties to the environment, several forms of representation of decision makers' preferences over problem solution alternatives are presented, through quantitative and qualitative models, these last ones based on the fuzzy set theory. To make possible the processing information in different structures, transformation functions are used to reduce those structures to a unique base, that is, fuzzy preference relationships. The preparation of information for decision making is accomplished in two stages: application of methods for aggregating preferences and ordering solution alternatives. Situations for individual and group multicriteria decision making are considered. For group situations, in particular, consensus schemes are presented, with prominence for the participation of a moderator in supervised processes. Enlarging the perspective of the best alternative choice, a methodology is proposed for analysis of solution quality through the concept of harmonious solutions. In this sense, a study is accomplished on combinations of aggregation and ordering operators, with a posterior analysis of the results, with distinction for the use of min operator for preferences aggregation and max operator for alternatives ordering. As a scenario for this study, the practice of decision making in the strategic management of electric energy companies, in particular of the Minas Gerais State Energy Company, has been used. Finally, the work presents a computational prototype, in which the developed techniques, approaches, models and methods are implemented.