A Hybrid MBO-VND algorithm to solve distribution network reconfiguration problem

The use of bio-inspired meta-heuristics in electrical engineering optimization problems is subject of a considerable number of studies, specially in distribution network reconfiguration (DNR). Such approaches shows to be of easy implementation and understanding, presenting, majorly, good results. As systems grows in scale, the difficulty to find the best solution also increases, thus needing modifications in the original structure of such algorithms to improve the search. One way of doing that is through the use of a local search procedure using a second technique such as greedy adaptive search procedure (GRASP) and variable neighborhood search (VNS). These latter meta-heuristics normally needs a good starting point to reach the best result for the majority of systems. As bio-inspired meta-heuristics usually presents better results in comparison with auxiliary algorithms (e.g. Prim) normally used to find the starting points, such techniques can be used in replacement of it. In this context, this paper proposes a hybrid approach to solve DNR aiming losses reduction. The approach, named hybrid monarch butterfly optimizer variable neighborhood descent (HYMBOV), is based in the monarch butterfly optimization (MBO) and variable neighborhood descent (VND). In HYMBOV, MBO is used to search the best solution using a small number of iteration and, after that, VND seek for improvements in the solution with the MBO solution serving as the starting point. All the electrical analysis is performed through OpenDSS. The algorithm was fully discussed when tested in two well known systems of the literature: 33- and 84-bus.