This work presents the formulation and the computational implementation of an evolutionary algorithm (EA) to solve the reconfiguration problem of radial distribution systems under different loading condition. In a network reconfiguration problem, the string used for representation could have information on system topology. The codification adopted uses the decimal representation model. The mutation, crossover and selection operators for the reconfiguration algorithms are analyzed. The selection approached used are stochastic tournament, elitist and a mixed techniques that seeks to use the advantage of the stochastic tournament and elitist crossover operators. The crossover operator was developed considering the representation structure of the chromosome that maps the linkage branches and the system radiality exigency, and another structure that considers the topology and the network operation feasibility for the effectuate the genetic material change. Strategies to adequate the parameters and improve the efficiency of the evolutionary algorithm are proposed , emphasizing the dynamic variations of the crossover and mutation rates. The initial population topologies are randomly generated considering the radiality constraints and the fast constructive algorithms of minimal generation trees of Prim and Kruscal. The obtained results are analyzed and presented for two real systems, one of 135 buses and another of 215 buses.
Distribution networks; Losses optimization; Combinatorial optimization
