Optimal Management of Electric Vehicle Interactions with Power Grid Considering Battery Degradation

: (Alternative Format Thesis)

Abstract

Modern energy systems are facing increasing pressures in achieving net-zero targets, caused by the combination of increasing intermittent renewable generation, low-carbon demand, and the penetration of electric vehicles (EVs). In recent years, vehicle-to-grid (V2G) is gaining popularity, which schedules the charging and discharging time, power, and even locations of grid-connected EVs (GEVs) for improving the stability and economy of the power grid. With V2G, GEVs can be used as distributed energy storage to absorb excess renewable generation and provide ancillary services to the grid. To capitalize on the benefits, GEVs' energy storage capacity should be properly scheduled, and battery aging should be effectively mitigated when providing V2G. Existing V2G behaviour management methods to some extent fail in realizing battery anti-aging scheduling, absorbing renewable energy, and algorithm real-time deployment.

This thesis focuses on modelling and mitigating battery degradation when providing V2G services to the power grid with renewable generation. The main achievements of this research can be summarized as follows:
(1) It establishes a series of battery aging model by analysing the impact of aging cycles under different depths of discharges (DoDs), charging and discharging rates (Crates), and state of charges (SoC) under various working conditions. By comprehensively considering battery aging characteristics, it can generate a more accurate battery life loss quantification feedback cost signal for guiding vehicle battery management in V2G services.
(2) It proposes a novel active battery anti-aging EVs grid integration scheme to mitigate battery aging costs caused by V2G services. EVs charging management is established as a multi-objective mathematical optimization model, where the reduction of battery degradation is designed as one of the optimization objectives to realize battery protective V2G scheduling. Meanwhile, an improved heuristic algorithm is developed to enhance the computational efficiency of the active battery anti-aging V2G scheme. Compared to the conventional V2G management method, the battery aging cycles are reduced by 48.9%, which indicates that the battery degradation phenomenon during the V2G application is suppressed effectively.
(3) It develops an online deployment method of the battery aging model to realize practical battery management in V2G services. An analysable battery degradation cost quantification method is proposed to provide a single-step battery life loss feedback signal for facilitating anti-aging V2G management. By deploying the proposed method in energy management optimization problems, battery operation can be online scheduled while significantly reducing aging costs.
(4) It for the first time integrate battery protection mechanism in V2G scheduling in the microgrid with renewable energy penetration. It develops a novel mixed-learning framework to ensure V2G scheduling optimality and real-time performance. The enhanced frequency response and dynamic extreme learning machine approaches are used to schedule the charging behaviour of EVs based on the real-time sampled grid frequency state information. It ensures scheduling optimality and real-time performance at the same time in V2G in renewable energy systems.
(5) Finally, this thesis presents a model-free solution for flexible GEVs charging management. It designs a novel battery protective V2G behaviour management framework based on a deep reinforcement learning method. The developed multi-target learning method can derive the optimal GEVs charging management strategy by better balancing the cost-benefit effectiveness between different factors.

This study will greatly contribute to accurate and computationally efficient battery aging quantification and battery anti-aging GEVs charging management. The derived V2G schemes serve as an economic and reliable solution for GEVs charging management in energy-transportation nexus with renewable energy penetration.

Date of Award	26 Apr 2023
Original language	English
Awarding Institution	Department of Electronic & Electrical Engineering
Supervisor	Chenghong Gu (Supervisor) & Jun Zang (Supervisor)