Towards the improvement of multi-objective evolutionary algorithms for service restoration

Distribution systems (DS) service restoration is a multi-objective, multi-constraint, combinatorial and non-linear optimization problem that must be quickly solved. Four multiobjective evolutionary algorithms (MOEAs) are proposed, which combine prominent aspects of highlighted MOEAs in the literature, for dealing with SR problem. Their main differentials are the providing of improved Pareto fronts and prioritization of switching operation in remotely controlled switches, which is widely used in smart grids. Proposed MOEAs were compared with a MOEA from literature by several tests in a large-scale DS. The MOEAs' performance was measured by four metrics for evaluation of Pareto fronts and Welch's t-hypothesis test was used for statistical comparison of such performance. Test results indicate all proposed MOEAs performed better than the MOEA from literature.