| Author: | Su, Chunyuan |
| Title: | Energy management strategy for fuel cell hybrid vehicles based on ECMS and DDPG algorithms |
| Degree: | M.Sc. |
| Year: | 2026 |
| Department: | Department of Electrical and Electronic Engineering |
| Pages: | vi, 58 pages : color illustrations |
| Language: | English |
| Abstract: | In recent years, data from reputable international organizations indicate that the global number of automobiles has experienced steady growth, increasing by an average of 5% annually over the past ten years. By 2023, this figure had surpassed 1.5 billion. Among these, conventional internal combustion engine vehicles remain dominant, accounting for roughly 40% of the world's annual oil consumption. These vehicles release substantial quantities of pollutants into the atmosphere during operation, contributing to billions of tons of carbon dioxide emissions each year, along with numerous harmful byproducts such as nitrogen oxides and fine particulate matter. Consequently, the global average temperature has risen by approximately 1.1℃ over the last hundred years, while sea levels have been rising at a rate of about 3.6 millimetres per year. The growing challenges related to energy consumption and environmental degradation have increasingly constrained global progress and sustainability efforts. [1] In the realm of new EVs, there are primarily three types: FEV, hybrid, and FCs. Fully electric vehicles demonstrate clear advantages for short-distance urban transportation. For instance, in China, the adoption rate of battery electric vehicles for urban commuting exceeded 60% in 2022, while the current average energy density of batteries remains within the range of 150−200 Wh/kg. Fuel cell vehicles use hydrogen as their energy source, and their reaction products are clean. Theoretically, each kilogram of hydrogen can provide approximately 33.6 kWh of energy, far exceeding the energy density of traditional fuel. However, its research and development costs are extremely high. The typical research and development expenditure for FCVs is approximately two to three times higher than that of conventional fuel-powered vehicles. By 2023, the global number of hydrogen refuelling stations had reached only around 800, with most of them located in regions such as Europe, North America, and Japan. Additionally, the construction cost of a single hydrogen refuelling station can go up to one to two million U.S. dollars. Challenges remain, including the limited durability of fuel cells, the need for enhanced safety measures, and an overall slow pace of commercialization. Hybrid vehicles integrate the strengths of both systems. Currently, they primarily rely on the coordinated operation of an internal combustion engine and an electric motor, dynamically switching between power sources based on varying road conditions to efficiently balance energy usage and performance. Fuel cell hybrid vehicles offer even greater uniqueness in this regard. By integrating fuel cells and power batteries in an innovative manner, their energy management strategy plays a vital role in handling complicated road conditions. Specifically, the approach focused on minimizing equivalent energy consumption dynamically adjusts the energy output of the dual battery system based on real-time factors such as road conditions, vehicle load, and remaining energy, aiming to reduce overall energy usage while maintaining sufficient power. This capability is essential in determining the system's competitive edge. EMSs play a role in efficiently regulating the power system's energy output and minimizing energy waste. In the context of hybrid vehicles, these strategies manage the energy distribution between the engine and the motor by employing specific control mechanisms. Depending on the driving conditions, the vehicle can be powered solely by the battery and motor or by a combination of both, maintaining its original performance while enhancing overall efficiency. Accordingly, this thesis thoroughly examines the energy management strategies within power systems. Initially, a single-axle parallel hybrid vehicle model is constructed. This thesis explores the structural classifications of hybrid power systems from two perspectives: the transmission method within the power system and the extent of hybridization. Emphasis is placed on examining the structure and operational principles of single-axle parallel hybrid vehicles. By employing an experimental modelling methodology, simulation models of the rear-mounted hybrid powertrain system—comprising the vehicle, engine, motor, and power battery modules—are constructed via cruise control functionality, with defined signal flow interactions among the constituent modules. Within the MATLAB/Simulink environment, a rule-basedλ energy management scheme is devised, and model is validated experimentally to confirm its dependability, forming a solid basis for subsequent enhancement of the energy management approach. Following this, an energy management technique emphasizing the minimization of equivalent fuel usage is proposed. Since the rule-based method does not fully harness the fuel economy potential of hybrid vehicles, an optimized strategy centered on equivalent fuel consumption reduction is formulated. This marks a transition from global optimization to instantaneous optimization at the extremum level, substantially improving fuel economy performance. Simulations are conducted under diverse operational conditions to identify motor operating regions, analyze energy distribution among various power components, and calculate the associated equivalent coefficients. The results reveal that, under the same simulation setup, the EMS based on minimizing equivalent fuel consumption achieves greater fuel savings than the rule-based approach. Moreover, under different simulation scenarios and environmental conditions, the extent of fuel efficiency improvement varies accordingly. Finally, an optimized energy management strategy based on learning algorithms was developed to minimize equivalent fuel consumption. Across various operating conditions, the initial values of the equivalent factor in the minimum equivalent fuel consumption strategy differ significantly, highlighting a strong correlation between the initial equivalent factor and the specific operating conditions. In addition to this strategy, a reinforcement learning algorithm was trained within a continuous action space to evaluate the performance of the condition recognition system. Simulations were carried out under diverse operating scenarios. The results demonstrate that, under identical simulation conditions, the learning-enhanced minimum equivalent fuel consumption strategy achieves a notably better fuel efficiency compared to the conventional minimum equivalent fuel consumption approach. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 8886.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 2.21 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/14457