Knowledge Assisted Deep Reinforcement Learning for Electric Vehicle Charging Control

Deep reinforcement learning (DRL) is a promising data-driven approach to solve the electric vehicle (EV) charging problem. However, DRL based charging strategies maybe not always meet the requirements of users. In this paper, a knowledge-assisted algorithm combining twin delayed deep deterministic p...

Full description

Saved in:
Bibliographic Details
Published in:2022 IEEE 6th Conference on Energy Internet and Energy System Integration (EI2) pp. 1882 - 1887
Main Authors: Zai, Rui, Guo, Ye, Liu, Qiong, Sun, Hongbin, Wu, Qiuwei, Xiao, Li
Format: Conference Proceeding
Language:English
Published: IEEE 11.11.2022
Subjects:
Charging control
Costs
Deep learning
Electric vehicle charging
imitation learning
Process control
Reinforcement learning
Simulation
System integration
twin delayed deep deterministic policy gradient algorithm (TD3)
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Deep reinforcement learning (DRL) is a promising data-driven approach to solve the electric vehicle (EV) charging problem. However, DRL based charging strategies maybe not always meet the requirements of users. In this paper, a knowledge-assisted algorithm combining twin delayed deep deterministic policy gradient (TD3) algorithm and imitation learning is proposed to control the charging process of EVs. The purpose is to minimize the cost while charging to the desired value. With the assistance of knowledge, the out-of-limit actions are corrected, so only the reward function to constrain the cost needs to be set, speeding up the convergence speed of the algorithm. To make the actor network acquire the capability of charging to the desired value, imitation learning is used to correct the actor network. The simulation results demonstrate the superiority of the knowledge-assisted TD3 algorithm with imitation learning.
DOI:10.1109/EI256261.2022.10116271