With the aim of relocating consumption and production, energy management at the scale of a micro-grid seems to be a lever to improve system efficiency, lifespan and profitability. The purpose of our work is to implement an energy management policy using reinforcement learning techniques in order to maximize the profits and considering different time-varying phenomena: variable power production from a PV system, variable power consumption from a DC load and variable energy price. This work will also take into account the battery degradation as a cost to be minimized. In the near future, this application would enable a better integration of electric vehicles in the grid.