EV-HVAC System Joint Control using Reinforcement Learning

Electric Vehicles has been evolved as keystone in the transportation system. In the urban areas buildings offer as the critical charging infrastructure for EVs. However, EVs have the advantage of the flexibility of the demand and can be a buffer source while discharging. Joint control of EV and HVAC will help in reducing the total power bill of the building during differential pricing of electricity throughout the day. EV-HVAC system joint control problem is considered while maintaining the HVAC systems thermal condition and the desired SoC of EV using Reinforcement Learning. Q learning algorithm is used in decision making. It is simulated in a designed environment consisting of Building HVAC system, EV and Grid. The rewards for the episodes were plotted and Hyperparameter tuning is analyzed. The results showing State Transitions table representing the correct actions with rewards and incorrect actions with penalty was obtained. The Q table represented the right actions for each state with highest reward.