Feasible Domain-Driven Genetic Operators for Distribution Systems Optimal Reconfiguration

In this paper, the problem of minimum losses reconfiguration in weakly meshed medium voltage (MV) distribution systems is dealt with. The presence of meshes can either be a cause of the presence of particular loads, favoured customer (FC) nodes, requiring a double supply path or of distributed generation (DG) nodes, that cannot work in islanded mode. The problem has been brought back to the more general problem of radial networks reconfiguration with the major aim of minimising power losses subject to constraints on voltage drops at load nodes and current in branches as well as on the partition of power generated among supply nodes in the network. The problem is non-linear and combinatorial considering the objective function, the constraints and the unknowns; for its solution, the authors have used a genetic algorithm. For the creation of the first generation as well as of the subsequent generations, a set of new generation and perturbation operators have been set up. They guarantee that all solutions represent a radial structure while supplying all the loads. In this way, the generation of unfeasible solutions is avoided and the verification step afterwards is no longer necessary. The applications show that such operators allow a wider exploration of the search space, while their implementation is quite easy and does not require a particular computational effort.