Analysis of Solution Quality of Multiobjective Evolutionary Algorithms for Service Restoration in Distribution Systems

Abstract

This paper presents a comparative study of six multiobjective evolutionary algorithms (MOEAs) with the node-depth encoding (NDE) which have been used to solve the service restoration in distribution systems. The study has been divided into three steps: (1) the MOEAs have been evaluated taking into account the switching operations necessary to find adequate restoration plans considering multiple nonlinear constraints and objective functions; (2) the MOEAs have been employed to solve four different datasets with 3860, 7720, 15,440 and 30,880 buses, respectively; (3) comparisons have been performed using the hypervolume indicator and the results obtained with each approach are statistically compared using Kruskal–Wallis nonparametric tests and multiple comparisons. In addition, this paper provides a comprehensive evaluation of six combinations of MOEAs based on NDE and our objective is to identify the features of each approach that consistently produce best results applied to network reconfiguration for service restoration in distribution systems. Simulations results have shown that MOEA based on NDE with crowding distance and strength pareto found good configurations with low switching operations and explored the search space better than others approaches used in this paper, approximating better the pareto-optimal front.