Energy Management Of V2G Systems And Application Research In Smart Grid

Environmental pollution and fossil fuel shortage pose a threat to sound development of human society. Nowadays, energy saving, emission reduction and renewable energy have become trends of global economic development. Smart grid is the next generation of power systems, which can integrate large amount of distributed renewable energy resources based on strong power grids and reliable communications networks. It also collects power information over the whole power system to achieve the optimization and self healing of the system and make the system greener, more efficient and more secure. Electric vehicles use electrical energy as power sources with zero discharge, which can implement energy saving and environmental protection. With penetration of electric vehicles, their charging loads will significantly impact power grids, causing voltage drop, overload, harmful harmonics, etc. Thus, it is very necessary to implement coordinated control on charging and discharging behaviors of electric vehicles. V2G (Vehicle-to-grid) systems are able to manage a large number of vehicles’batteries and control their charging and discharging processes via energy scheduling or dynamic pricing. In V2G systems, vehicles’energy demands are satisfied while batteries are used to achieve peak shaving and valley filling. V2G systems achieve information and energy interactions between electric vehicles and power grids and improve flexibility and interoperation of power systems.This thesis mainly studies energy management problems of V2G systems. At first, existing energy scheduling methods and demand response management methods of V2G systems are introduced. Considering the disadvantages of the existing V2G energy management methods, this thesis proposes new energy management methods and system models. The contributions of the thesis are as follows:Firstly, by considering unidirectional V2G charging energy scheduling in single V2G system, an entropy based fair energy scheduling method is proposed. The entropy based fairness considers three metrics in energy scheduling at the same time, and it uses entropy weight method to weight the three metrics. For the dynamic problem of energy scheduling, neural dynamic programming is used to obtain an approximate optimal scheduling policy. Numerical results show that the proposed method can reduce the load peak in a grid and ensure the performances of the metrics.Secondly, by considering bidirectional V2G charging and discharging energy scheduling, a contribution based fair energy scheduling method is proposed. In the contribution based fairness, electric vehicles with higher contribution values will have higher priorities to obtain charging energy. The contribution values are determined by energy and timing of charging and discharging. Adaptive dynamic programming is employed to solve the dynamic problem of energy scheduling. Numerical results show that the proposed method can achieve peak shaving and valley filling, and it also achieves quick recovery of vehicles’batteries that have deeply discharged and ensures fairness of fully charging time.Thirdly, by considering the influence of electric vehicle mobility on multiple V2G systems, a model of V2G mobile energy networks is proposed. In the model, each district node has a V2G system, and demand response dynamics of systems are coupled through vehicle fleets. Synchronization of complex networks is used to analyze dynamic behaviors of V2G mobile energy networks. The analysis shows that symmetrical fleets can achieve a synchronous state of a network and in turn balance power demands among districts. Real vehicle and power grid data are adopted in simulations to show the performances of the proposed model in a real scenario.Fourthly, based on V2G mobile energy networks, a robust demand response algorithm for asymmetrical fleets is designed. A difference equation system is constructed to describe demand response dynamics under a case of coupled nodes. The robustness of the demand response algorithm is analyzed. Two real vehicle travel datesets are used to analyze statistical features of V2G mobile energy networks driven by the real data and the interactions between the real vehicle fleets and the proposed demand response mechanism.Finally, the last chapter concludes the researches in the thesis and evaluates the research results. Future work on V2G systems is also given.