Heurística construtiva em dois níveis visando a determinação das decisões de investimento no planejamento da expansão de sistemas de transmissão

Abstract

The transmission network expansion planning of electric energy consists of determining the reinforcements to be built, within a pre-defined set of reinforcements that are candidates for expansion, for supply the demand for a power system, in order to minimize investment and operating costs in a planning horizon. This optimization problem is difficult to solve because it presents some particularities: (i) several possible solutions, which leads most of the algorithms to converge to local optimal solutions; (ii) mixed-integer and combinatorial nature of investment options; (iii) large size of transmission systems, referring to real systems; and (iv) existence of non-connected electrical systems. These particularities refer to the main difficulties in the development of efficient algorithms. The development of algorithms/techniques capable of reconciling the achievement of quality solutions, regardless of the size of the system, and low computational effort is an attractive field of research and development. Therefore, in this research, a new constructive heuristic algorithm is proposed. This new technique, besides maintaining all the advantages inherent to this class of algorithms, differs from the constructive heuristics in the literature by: (i) presenting two-stage of search, these being provided by the adopted representation (hyperbolic tangent function and variations of slope) for decision variables. The two-stage search strategy allows for a better exploration of the solution region and, consequently, greater assertiveness in the last gotten solution; (ii) Use an unprecedented performance indicator to specify the expansions to be carried out in the electricity transmission system. Analyzing the results found, the proposed heuristic strategy brings an interesting contribution to the area, since it has managed, for the analyzed systems, to meet two important requirements and pursued by most of the methodologies used in solving the problem addressed here that are the low computational effort and the quality of the final gotten solutions.