Research On Distributed Vehicle-to-grid Scheduling Strategy

The ever-increasing global energy crisis and the increasing public awareness of environmental protection have promoted the rapid development of electric vehicles(EVs).As a new type of power load,how to coordinate large-scale EVs to the power grid has attracted extensive attention.Statistically,private vehicles spend only 4%of their day in traffic,making it possible for electric private vehicles to be used for other purposes during the remaining 96% of their free time.The vehicle-to-grid(V2G)technology can optimize the idle EVs to charge during the grid load valley period and feeding the grid as a distributed power source during the load peak period.This bidirectional energy flow technology builds a bridge between the power grid and EVs,and its optimization decision variable are the charging and discharging power of the EV at each time slot.However,with the increase of the scale of EVs,the variables that need to be optimized during the unified dispatching process of the traditional centralized V2G become larger,and the computa tional complexity becomes higher,making the problem difficult to solve.Additionally,this also raises privacy anxiety among EV users,as EV aggregator need collect charging information from individual EVs.Therefore,high-efficiency,high-safety,and high-stability scheduling strategies have become one of the most innovative and challenging research directions in the field of V2G technology.Based on the technical requirements of high computing performance and high information security in V2G,this paper systematically studies several key issues such as the architecture,scheduling strategy,and multi-dimensional verification of distributed V2G,and obtains some innovative research results.(1)In order to achieve the V2G of high-efficiency,high-safety,and high-stability,various existing scheduling models and strategies are compared and analyzed,and on this basis,an innovative distributed internet of smart charging points(ISCP)architecture is proposed.First,a comprehensive review of the V2G from multiple perspectives such as computing performance,communication performance and information security performance was conducted,which can provide beneficial technical reference for peers concerned about large-scale EVs access to the power grid.Second,based on the above performance analysis,a distributed V2G scheduling architecture is proposed,i.e.,ISCP.The core of the architecture is the EV charging point that integrates low-cost computing module,namely,intelligent charging point,which has the function of local computing and local information processing.In addition,intelligent charging points coordinate with each other through distributed scheduling,which creates favorable conditions to reduce the computational complexity of the model,avoid user privacy disclosure,and optimize the power energy flow.(2)In order to take into account the scheduling effect and computing efficiency of the V2G scheduling strategy,based on the proposed ISCP architecture,two demand response strategies with high computational performance,namely day-ahead load peak-shaving and valley-filling,real-time load flatting and photovoltaic absorption,were designed and formed.Specifically,the V2G is divided into six typical scenarios to form the calculation model of “single vehicle,single point,single problem”,which can efficiently coordinate the random access of large-scale EVs and seamless absorption of renewable energy.The simulation results based on the campus distribution grid of Southern University of Science and Technology verifies the correctness and effectiveness of the two proposed demand response strategies,as well as their ability to adapt to the requirements of distributed scheduling such as “scalability” and “high computing performance”.(3)In order to meet the strong demand for information security of all participants in the V2G,a secure interaction scheme for the whole chain digital assets based on ISCP architecture is proposed.On the one hand,a V2G dataset is generated for commercial centers,including conventional load data,photovoltaic output data and EV charging data.On the other hand,based on real datasets,two methods of user privacy protection and data support empowerment based on long and short term memory network and digital asset protection based on deep learning and federated learning are proposed respectively.Specifically,distributed edge computing is used for scheduling prediction at charging point to protect user privacy;desensitized data is used for model training through deep learni ng at charging stations,which can avoid the challenge of difficult future data acquisition;model integration is performed at cloud servers through federated learning to protect the digital assets of charging stations.The simulation results verify the correctness and effectiveness of the two proposed data-driven methods,as well as their ability to adapt to the requirements of information security such as “burning after scheduling” and “desensitization training”.(4)In order to deeply analyze the coupling mechanism among key elements of V2G,a multi-dimensional model scheme based on decentralized internet of smart charging points architecture was creatively established.On the one hand,a cyber-physical-energy joint model has been developed.In terms of energy optimization,a distributed multi-layer parallel V2G model is used,which can ensure the scheduling effect and improve computing efficiency;in terms of network communication,a small world network can be used to ensure communication efficiency and reduce wiring cost.On the other hand,a multi-time scale V2G framework of “two-level coordination,situational coupling,and detailed step by step” is proposed.The traditional optimization method is used for day-ahead energy scheduling to obtain the safety boundary in real-time scheduling;the real-time rolling energy dispatching uses the spatial decoupling alternating direction method of multipliers to decouple the connection between charging points to improve dispatching efficiency and achieve global convergence.The simulation platform and physical platform results verify the correctness and effectiveness of the proposed framework in energy scheduling,privacy protection,and network communication,as well as their ability to adapt to the requirements of distributed cyber-physical-energy system such as “network robustness” and “computation decoupling”.