Assessment of optimal distribution network reconfiguration results using stochastic dominance concepts

In the recent research on power and energy system optimization, many deterministic and heuristic solvers have been proposed. Each proposal claims that the new solver is better than the previous ones on the basis of performance indicators, which are often limited to the best solution found or to simple statistics (mean, median, standard deviation). This paper introduces a new and more significant performance indicator based on the concept of first-order stochastic dominance. This indicator can generally compare the solutions of a given optimization problem for which the global optimum is not known. The optimal discrete distribution network reconfiguration for a real-scale system was taken as an example problem, to show the characteristics of the proposed indicator. The results obtained show the effectiveness of the proposed indicator to limit the acceptability of better solvers to the ones that actually exhibit enhanced performance with respect to incrementally improved benchmarks. (C) 2016 Elsevier Ltd. All rights reserved.