EV Range Management by Optimal Control of HVAC using Reinforcement Learning

Electric Vehicles and Battery-Fuel_Cell hybrid vehicles are increasingly becoming popular in the market, especially in the commercial vehicle segment. Range estimation and control is of paramount importance as it is the main cause of anxiety among the vehicle owners. This paper discusses application of Reinforcement Learning (RL) to achieve range control. In RL, the learning agent choses actions dependent on the state of the environment and gets a reward in return. Ultimately the agent will learn the policy of choosing the actions for each state such that his long-term reward is maximized. The technique of RL has been applied for various scenarios where in a look up table (between the states of a system and actions to be taken) needs to be developed for optimal performance. In this paper, we use RL to manipulate other energy sources and sinks like Fuel Cell and HVAC (in addition to the battery which is the main energy source) for range control, and thereby achieve the optimal performance as by the system experts.