Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Electrical and Electronic Engineering | en_US |
| dc.contributor.advisor | Bu, Siqi (EEE) | en_US |
| dc.creator | Huang, Yihang | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/14144 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Secure and economic operation of power and transport systems : a reinforcement learning-based approach | en_US |
| dcterms.abstract | As the continuous development of electric vehicle technology and the comprehensive promotion of carbon neutrality work, electric vehicles have great advantages in reducing environmental pollution and realizing optimal allocation of energy resources due to their characteristics of low pollution, zero emission and high energy efficiency, and are of great significance in promoting the energy revolution and realizing the dual-carbon goal, receiving worldwide attention. Began to be widely used in the world. Governments began to take electric vehicles as an important way of the transformation of the energy industry structure, experts from various countries have made many technical breakthroughs in the research of EVs, making EVs in the performance and economy has been greatly improved, electric vehicles are beginning to enter the era of promotion and application worldwide. However, under the influence of a lot of elements, for instance, the travel behavior of electric vehicle users, life rules and external environment, the charging behavior of EVs is uncertain in time and space, and the randomness of time distribution mainly due to charging duration and initial charging time. The randomness of spatial distribution mainly refers to the randomness of vehicle charging location. As the continuous grow of EVs ownership, the charging without coordination of a lot of EVs will increase the voltage deviation and network loss of the grid, and affecting the economic operation of the power system. Therefore, it is very important to optimize the control of coordinated optimized charging of EVs. Reinforcement learning can adjust and optimize strategies in real time to cope with these random changes, break the limitations of traditional charging strategies, gain experience through continuous training and learning, get the optimal control strategy, achieve better control effects, and further reduce the adverse effects of charging without coordination. | en_US |
| dcterms.abstract | This paper reviews the current research methods and status of control strategies for charging EVs to the grid. On the one hand, it describes that most current studies are related to the analysis of electricity market, few papers are analyzed from the impact on the grid side. Therefore, it shows that this paper has research value to analyze the negative influence of charging without coordination of EVs on the power grid from the perspective of the power system. On the other hand, it shows that the charging act based on EVs is stochastic in time and space, and the traditional research strategies and methods of electric vehicles have certain limitations. In summary , this paper builds reinforcement learning SAC algorithm to study the influence of coordinated charging of EVs on the grid. The reinforcement learning algorithm used here is SAC algorithm, which has the maximum entropy reinforcement learning frame. It encourages the algorithm to explore by introducing the entropy regularization term. While trying to maximize the expected reward, it also maximizes the strategy entropy, thus improving the performance of the algorithm with strong exploration ability and high robustness. The coordinated charging of EVs can be effectively used to decrease the negative influence of charging on the power grid. | en_US |
| dcterms.abstract | Combined with the working principle and situation of EV charging, this paper starts from the random model of EV access to the distribution system, and analyzes the influence of electric vehicle access on the grid according to the charging situation of electric vehicle under disordered control. Establishing a model of EV charging load, and the influence of unordered charging on the EV grid under different permeability levels is analyzed from two aspects: voltage offset and network loss. On this basis, a control method of unidirectional coordinated charging is proposed by establishing the SAC algorithm model of reinforcement learning. At the same time, combined with V2G technology, the charge/discharge of EVs are optimized. | en_US |
| dcterms.abstract | The main research result of this paper is to take reinforcement learning model to coordinated control EVs charging/discharging. Based on the purpose of reducing the adverse influence on charging EVs in the state without coordination on the power system, proposing a unidirectional coordinated charging method for EVs combined with reinforcement learning SAC algorithm. This method aims to reduce the voltage offset and network loss, and controls the reactive power capacity Q of charging piles by generating action values at corresponding times, to control the voltage. Through simulation analysis, the charging effect of EVs before and after adopting the optimized approach is compared.At the same time, the possibility of voltage exceeding the limit is also considered in the optimization control of unidirectional coordinated charging. In this chapter, adopting the strategy of active power reduction, to further optimize the voltage offset and improve the charging performance of EVs. | en_US |
| dcterms.abstract | Finally, combined with the SAC algorithm of reinforcement learning, proposing a bidirectional coordinated charging/discharging control method for EVs. Combined with the aggregated V2G model, the charging pile has two control action values, one control reactive power and one control active power (only during discharging). Though controlling the reactive power and active power capacity of the charging pile, the voltage regulation is realized, and the adverse effects include voltage deviation and network loss are optimized and improved. The charging performance of EVs under bidirectional coordinated and charging without coordination is analyzed by simulation, and further validates the feasibility of the proposed method. | en_US |
| dcterms.extent | 76 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2024 | en_US |
| dcterms.educationalLevel | M.Sc. | en_US |
| dcterms.educationalLevel | All Master | en_US |
| dcterms.accessRights | restricted access | en_US |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 8480.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 3.11 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/200/14144