Transactive Electric Vehicle Agent: A Deep Reinforcement Learning Approach
Swastik Sharma, Swathi Battula, Sri Niwas Singh · 2024
Bid-based Transactive Energy System (TES) designs for distribution systems have several merits, including fair compensation of participating Distributed Energy Resources (DERs) where they can voice their preferences using bids/offers. In this research, a Deep Reinforcement Learning (DRL)-based Proximal Policy Optimization (PPO) algorithm with recurrent neural networks has been employed to derive bids/offers from participating Electric Vehicle (EV) agents by modelling the problem as a Partially Observable Markov Decision Process (POMDP). This helps in preserving the non-linear characteristics of the problem, i.e. obtaining DERs’ price-sensitive bids/offers while considering user goals and constraints, owing to the highly efficient function approximation characteristics of Deep Neural Networks (DNNs) employed in DRL-based algorithms. The obtained results demonstrate the convergence of the agent’s policy in real-world data having dynamic price fluctuations.