Energy management programming to reduce distribution network operating costs in the presence of electric vehicles and renewable energy sources

Due to growing environmental and economic constraints, countries are exploring renewable sources such as wind, solar, and fuel cells to save energy, and develop the use of dispersed generations. Thus, the use of electric vehicles (EVs) is on the rise. On a large scale, either of these technologies can have damaging effects on the electricity grid; however, with suitable consumption-side management and programming, technologies and energy storage resources can reduce these effects. Thus, energy management optimization has become an interesting topic of research. Accordingly, the effect of the integrated aggregation of PEVs to the grid for the charge/discharge process and the resulting grid instability, especially at load peak time, is the main challenge to the use of these vehicles. The contribution of this paper includes the presentation of a model for managing the coordinated and uncoordinated charging system of grid-connected EVs with wind power and photovoltaic power units as dispersed generation sources and dividing the EVs into 4 classes by considering the share of each in the grid and considering a random number of vehicles per class using the normal distribution function and implementing the incoordination in wind speed and solar irradiation. The proposed model uses a novel Reinforcement Learning (RL) based onDeep Q Network (DQN) algorithm to solve the multi-objective problem. In this model, the costs of annual energy losses and the operation of dispersed generation units are discussed in an integrated manner as the objective function. The simulation is performed on a 57-bus IEEE grid and the results show the efficiency and improved performance of the model.