Integrating relevant aspects of moeas to solve loss reduction problem in large-scale Distribution Systems

Distribution System (DS) reconfiguration for power loss reduction is usually formulated as a nonlinear, multi-objective and multi-constrained optimization problem. Recently an approach to solve this problem that presents a very good performance even for large-scale DSs was proposed. This approach, called MEAN, combines a multi-objective Evolutionary Algorithm (EA) based on subpopulation tables with a new tree encoding, named Node-Depth Encoding. In order to improve the performance of the MEAN, this paper proposes to incorporate elements from the Non-Dominated Sorting Genetic Algorithm-II (NSGA-II) into the MEAN. Basically, the proposed approach stores the non-dominated solutions from Pareto fronts (calculated as the NSGA-II does) into the MEAN by using additional subpopulation tables. The proposed method better explores the space of the objective functions and, consequently, better approximates the Pareto-optimal front. The direct result is the discovery of feasible solutions with significantly lower power losses and also a lower number of switching operations. Simulations results with DSs ranging from 632 to 1,277 switches have shown that with the incorporation of the additional subpopulation tables the performance of the MEAN is significantly improved.