Illuminating Metaheuristic Performance Using Vortex MAP-Elites for Risk-Based Energy Resource Management

With the current state of the electrical power system, regarding the increase of renewable generation integration and electric vehicle penetration to reduce gas emissions, the energy resource management problem becomes extremely complex to optimize to the significant dimensionality and uncertainty. Metaheuristic optimization algorithms become efficient methods since they guarantee a balance between optimal and practical solutions, but they lack explainability and are treated as black-box techniques. In this work, we introduce an improved version of the Multi-dimensional Archive of Phenotypic Elites (MAP-Elites) algorithm incorporating the Vortex Search to generate new candidate solutions in the iterative process. The VS MAP-Elites is then used to optimize the energy resource management problem for a 13-bus distribution network considering risk analysis due to the existence of extreme scenarios in the day-ahead operation. Two different behaviors of the problem were considered, namely demand response ratio and renewable ratio, and the effect that they have on metaheuristic performance was analyzed through the visualization of the elite archive. Results showed that VS MAP-Elites achieved better cost results compared to MAP-Elites, around a 25 % reduction, since it was able to diversify the search space finding better solutions for the considered problem characteristics.