Back to search

McMaster University

Deep Reinforcement Learning based Integrated Energy and Thermal Management System for Battery Electric Vehicles

Abstract

dc:description.abstract

This thesis explores the development of an Integrated Energy and Thermal Management System (IETMS) for Battery Electric Vehicles (BEVs) using Deep Reinforcement Learning (DRL). As electric vehicles become more prevalent, the need for intelligent control strategies that can manage both energy consumption and thermal demands grows increasingly important. Traditional energy management systems, such as rule-based or optimization-based approaches, often treat thermal and energy domains separately and lack the flexibility to adapt to real-time changes in driving and environmental conditions. To address these limitations, this work introduces a DRL-based framework that simultaneously considers thermal comfort, battery health, and energy efficiency in a unified control scheme. The system is developed within a detailed simulation environment in MATLAB/Simulink, which includes comprehensive models of the vehicle powertrain, thermal systems, and a battery aging model. The thermal side incorporates a dynamic HVAC model and cabin thermal dynamics, allowing the system to regulate cabin and battery temperatures under diverse environmental conditions. At the heart of the control system is a DRL agent trained using the Soft Actor-Critic (SAC) algorithm. The agent operates with a carefully defined state and action space and is guided by a reward function that balances passenger comfort, battery longevity, and energy savings. To enhance the agent's foresight, the control architecture also includes a velocity prediction module that combines Fuzzy C-Means (FCM) clustering with a Transformer-based model. This predictive capability allows the system to anticipate upcoming driving demands and proactively manage thermal loads and energy use. The proposed DRL-FCM controller is tested across a wide range of standard drive cycles and temperature conditions, including both hot and cold weather scenarios. Its performance is compared with conventional control strategies such as Proportional-Integral-Derivative controller (PID), Model Predictive Control (MPC), and standard DRL without velocity prediction. The results demonstrate that the integrated DRL-FCM system offers more adaptive, coordinated control, achieving better energy efficiency and improved battery health while maintaining thermal comfort for passengers. Overall, this thesis contributes to a novel solution for managing the complex energy and thermal dynamics of modern electric vehicles.

Degree

thesis:*
Department dc:contributor.department
Mechanical Engineering
Year dc:date.issued
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Jha, Anurag
Advisor dc:contributor.advisor
  • Emadi, Ali

Subjects

dc:subject × 9

Rights

Language dc:language.iso
en

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
McMaster University
Base URL
macsphere.mcmaster.ca/server/oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Jha, Anurag. Deep Reinforcement Learning based Integrated Energy and Thermal Management System for Battery Electric Vehicles. 2025. http://hdl.handle.net/11375/32160