Full metadata record
|dc.contributor||Department of Electrical Engineering||en_US|
|dc.contributor.advisor||Chan, K. W. Kevin (EE)||-|
|dc.publisher||Hong Kong Polytechnic University||-|
|dc.rights||All rights reserved||en_US|
|dc.title||Cooperative planning and multi-objective operation of electric vehicle charging stations||en_US|
|dcterms.abstract||In future smart grid, electric vehicles (EVs) would play a vital role to reduce air pollution and carbon emissions caused by conventional transportations while EV batteries could contribute to the power system dispatch as distributed energy storage devices. However, large-scale uncoordinated EV charging would bring challenges and difficulties to the control and operation of a power system due to the rapid growth of charging load demand, additional energy losses, deterioration of power quality, decrease of power grid economic efficiency, etc. Considering the infrastructure development of EV charging stations (CSes) as one of the key factors to the widespread use of EVs, this thesis conducts studies from the aspects of the planning and operation of EV CSes to meet the rapidly growing charging demand of EVs and eliminate any potential threats as a result. Because EV load prediction is the precondition of the planning and operation of CSes and distribution system (DS), the forecast methodology of EV demand is also one of the concerns of this thesis. The forecasting of EV load can be divided into long-term forecast and short-term forecast according to the time duration. For the long-term prediction, grey system forecasting theory model and nonlinear autoregressive (NAR) neural network model are firstly utilized in this thesis to forecast the annual growth in the number of EVs including electric buses (EBs) and non-EBs (including private electric cars, electric taxis, etc.). The effectiveness, rationality, precision and adaptability of the two models are evaluated and compared. Simulation results show that the NAR neutral network model has a better performance in long-term forecasting of EVs than the grey system forecasting model. Moreover, the deep belief network (DBN) method is firstly applied for accurate EV demand forecasting, the effectiveness of which is proved by comparing with other typical algorithms. For short-term forecasting, EV charging load is difficult to forecast accurately due to the non-stationary feature of traffic flow (TF) and erratic nature of charging procedures. In this thesis, TF is predicted by a novel deep learning based convolutional neural network (CNN) approach, and the model and data uncertainties are evaluated to formulate the prediction intervals (PIs). EVs' arrival rate is calculated based on the historical data and the proposed mixture model, and the EV charging process is studied by a novel probabilistic queuing model considering charging service limitations and drivers' behaviors. The proposed methods are assessed by using real TF data and the results demonstrate that the errors of the proposed method are reduced by about 30% compared to other widely-used approaches, and the probabilistic forecasting approach has better reliability and sharpness indices than other methods, which leads to high potential for practical use. According to the forecasting results, sufficient number of CSes should then be planned to meet the future charging demand of EVs while new feeders in the DS should be timely constructed to provide the required supply to the CSes. It is a common assumption in the present work that the CS planning and power system planning are managed by a single entity to carry out centrally, which is quite contrary to the reality. In fact, in the deregulated environment, DS and EV CS owners / operators are independent market participants and responsible for their own planning with different or even conflicting interests and objectives. Therefore, coordination of these interests and objectives is a critical and complicated problem for extensive integration of EVs in the liberalized market environment. In addition, the electricity market mechanism should be fully studied in the planning strategy making process. In this thesis, Nash bargaining theory is employed to formulate the cooperative planning for CSes and DS for the first time. A negotiated planning model of CSes and DS is established to achieve the most fair and Pareto-efficient payoff allocation for the two independent participants. Additionally, a novel locational marginal price (LMP) model to alleviate DS congestions with consideration of schedulable EV charging and flexible demands is proposed to model the real market environment while the deep belief network (DBN) method is firstly applied for the accurate forecasting of TF. Simulation results of a 38-node DS with high penetration of EVs and flexible demands have demonstrated that the realistic negotiated planning process and the consideration of DS market mechanism would improve the DS gain payoffs by 8.21% than those in the centralized plan, and the payoff gap between DS and CS is also reduced by 10.73%, which would boost the planning enthusiasm and lead to a more fair planning solution.||en_US|
|dcterms.abstract||Once CSes are constructed, the operation of CSes should be investigated to ensure the high efficiency and reliability of CSes and DS. In this thesis, the application potential of EVs in CSes are accounted in the electric power dispatch, especially with several conflicting and competing objectives such as providing vehicle-to-grid (V2G) service and coordinating with wind power. Further, to solve this firstly proposed highly constrained multi-objective optimization problem (MOOP) with the consideration of uncertainties of EVs and wind power, a decomposition based multiple group search optimization (MGSO/D) is proposed to efficiently reduce the computational complexity and innovatively incorporate the producer-scrounger model to effectively improve the diversity and spanning of the Pareto-optimal front (PF) while uncertainties are accounted by the estimation error punishment. The performance of MGSO/D and the effectiveness of the uncertainty model are investigated using the IEEE 30-bus and 118-bus system with wind farms and CSes. Four indices, namely convergence metric, span metric, spacing metric and lmax/lmin metric, are utilized to measure the solution quality of MGSO/D and other three well-established Pareto heuristic methods. The PF solutions obtained by the proposed MGSO/D in both small-size and large-scale cases show its superiority over other three algorithms on all 4 indices and demonstrate that it can propagate the search to obtain the uniformly distributed and diverse PF more effectively. Furthermore, a battery schedule framework is studied in this thesis to dispatch batteries between battery charging stations (BCS) and battery-swapping station (BSS) efficiently. Compared with the battery-swapping technology, fast charging technology has disadvantages that it takes a relatively long time and shortens the battery life much faster. The EV battery-swapping technology is a promising method to avoid the inconvenience of fast charging because of its flexibility. In this thesis, to improve the effectiveness of battery dispatch between BCS and BSS, an original two-direction battery dispatch mechanism to reduce the transportation cost are established and solved by the particle swarm optimization algorithm (PSO) method. The simulation results demonstrate that the optimized battery travel distance is reduced about 50% compared with the random travel. Moreover, considering the serving ability limitations, the K-means clustering algorithm is innovatively utilized to pre-partition the BCS and BSS to make the battery dispatch more efficient for the large-scale system, and the simulation results confirm the travel distance could be further shortened up to 15% by the pre-partition method.||en_US|
|dcterms.extent||xviii, 147 pages : color illustrations||en_US|
|dcterms.isPartOf||PolyU Electronic Theses||en_US|
|dcterms.LCSH||Hong Kong Polytechnic University -- Dissertations||en_US|
|dcterms.LCSH||Battery charging stations (Electric vehicles) -- Planning||en_US|
|dcterms.LCSH||Electric vehicles -- Power supply||en_US|
|dcterms.LCSH||Electric vehicles -- Batteries||en_US|
Files in This Item:
|991022208058003411.pdf||For All Users||4.55 MB||Adobe PDF||View/Open|
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: