The Brazilian Electricity Sector utilizes auctions as a way to contract energy and transmission lines. The winners of such auctions are the participants who offer the lowest tariff, aiming at economic efficiency in the contracting. A serious problem that has disrupted the start-up of some plants is the unavailability of lines to transport the energy from the generation center to the consumption center. Once the electricity sector is not vertical, the eventual lack of coordination between the construction of new power generation assets and new transmission lines results in power plants disconnected from the system. It is estimated a loss of around 8 billion reais between 2009 and 2013 (about 1,5 billion dollars in March of 2021), since the owner of the plant had right to the contracted revenue, even if there was no actual delivery of energy, unnecessarily raising the cost of energy and affecting the security of energy supply. In this way, changes were made in the auction's tender documents to avoid this problem. Aiming at another way of mitigating this problem, this research has the objective to propose a combinatorial auction of power generation assets and transmission lines in order to avoid the problem of system disconnection. In this model, both concessions would be bid together, avoiding the mismatch between them. As the Winner Determination Problem (WDP) can be very complex, since the number of possible packages grows exponentially with the increase of auctioned items, the focus of this thesis will be to solve the problem employing Evolutionary Algorithms (EA). EA are a class of algorithm that can obtain good solutions in a reasonable time but without the guarantee that the solution found is optimal. Thus, in this thesis was developed and proposed an algorithm named Constraint-Guide Evolutionary Algorithm (CGEA), which compared to other evolutionary algorithms presented in the literature, it seemed to be promising. The results obtained using the CGEA to find the solution of the auction proposed in this thesis were compared with solutions obtained when the methods of restricting allowed packages and divide the problem are applied so that the solution by exact methods can be employed. The proposed auction in this thesis together with the resolution by means of CGEA showed good results.