Hybrid Evolutionary Algorithm for Residential Demand Side Management with a Photovoltaic Panel and a Battery
Résumé
Residential demand side management (DSM) is one of the most challenging topics in smart grids. In this paper, a multiobjective model for the residential DSM over a 24-hour horizon is presented. This model consists of appliances, a battery and a photovoltaic panel. The resolution of this model is based on combining a multiobjective evolutionary algorithm (NSGA-II) and an exact solver (CPLEX). Solutions in this hybrid approach are incompletely represented, and optimally the exact solver determines the missing parts of the encoding. In our case, hybridization involves solving a MILP sub-problem by CPLEX to manage the battery and the photovoltaic panel constraints. Through case studies, It is shown that the coordination between the photovoltaic panel and the battery is effective to reduce the total electricity cost, the discomfort and the standard deviation of power consumed especially in summer conditions.