Research On Key Issues Of Electric Vehicle Charging Facilities Planning And Load Scheduling

The widespread application of electric vehicles(EVs)is one of the key methods to promote energy conservation and emission reduction in the low-carbon society.With the support of EV industry incentive policy and the breakthrough of EV’s energy storage technology,the number of EV has increased significantly.On the one hand,the large-scale charging behavior of EVs will cause a non-negligible and negative impact on the operation of the grid,such as aggravating the load peak-to-valley difference,generating harmonics,increasing network losses,etc.On the other hand,EV is a flexible and controllable load for grid.In the interactive environment of EVs and grid,reasonable planning of EV charging facilities and development of an orderly charging strategy for EVs can change the temporal and spatial distribution of EV loads,which will play a positive role in ensuring the safe and economic operation of the grid.Thus,this paper focuses on the key issues of EV charging facilities planning and EV loads scheduling.Starting from how to calculate the distribution of EV charging loads reasonably and accurately,this paper studies the dynamic expansion planning method of charging stations under the coupling of vehicle-traffic-grid,and then formulates a scheduling strategy that cooperates charging stations and sharing charging piles to match EVs.Lastly,the multi-objective scheduling strategy of EV considering user preference is formulated.Through above steps,the EV charging service can be optimized,the economic benefits of charging facilities can be improved,and the stable operation of grid can be maintained.The main works of this dissertation are summarized as follows:(1)The traditional EV load calculation method based on probabilistic modeldriven has problems such as low calculation efficiency,poor solution accuracy,etc.Based on this,this paper proposes a calculation method of data-driven EV load.This method comprehensively considers the EV battery characteristics,travel characteristics,charging characteristics and other factors to build an EV load calculation model.Then,the deep convolution neural network is used to perform feature extraction on EV user characteristics,and establishes a non-linear mapping relationship between EV user characteristics and travel characteristics to achieve high-precision calculation of EV loads that takes into account the characteristics of EV users.The proposed method effectively improves the efficiency and accuracy of EV load calculation,and avoids the problem that the traditional method needs to constantly update the EV load calculation model to make it complicated considering the difference of data characteristics.This EV load calculation method lays a theoretical foundation for the research of the grid to formulate EV scheduling strategies.(2)Considering the problem of interactive coupling between the power grid and the traffic network,the cost expenditure at each stage of the operation cycle,and the dynamic change of the EV penetration rate in the charging station planning,this paper proposes a dynamic planning method for the full life cycle of charging stations under the vehicle-traffic-grid coupling.The method selects the site and capacity of the charging station according to coupling relationship of the vehicle-traffic-grid to obtain the set of potential planning schemes for charging stations.Then,the paper builds a static planning model of charging stations based on the life cycle cost.On this basis,a dynamic planning model for charging stations that takes into account uncertainty of the EV growth rate is built.And the stochastic programming problem that takes into account uncertainty of the EV growth rate is transformed into a deterministic programming problem through scenario analysis.Finally,a quantum genetic algorithm based on quantum revolving gate is proposed to solve the programming model efficiently.The proposed method qualitatively distinguishes the cost components of charging stations and the time value of funds during the full life cycle on the basis of coupling and complex traffic network and power grid network,which improves economic benefits of charging stations obviously.The method also reflects the impact of the uncertain EV growth rate at each stage on the planning scheme,which solves the problem of dynamic expansion planning for centralized charging stations under the influence of uncertainty.(3)For the charging gap caused by the mismatch between the capacity of the charging station and the short-term charging demand of the EV after planning,an EV charging schedule strategy considering shared charging piles based on generalized Nash game is proposed.The strategy establishes a shared model of charging piles based on the generalized Nash game,which transforms the problem of shared capacity optimization into the problem of generalized Nash equilibrium that considers capacitycoupled constraints.This strategy proves the existence and uniqueness of the solution belonging to the generalized Nash equilibrium through variational inequality,and utilizes the smooth newton algorithm to obtain the optimal shared scheme of the charging pile when the generalized Nash equilibrium is reached.This strategy also builds a hierarchical scheduling model for the coordination of charging stations and shared charging piles,optimizing the charging time and location of each EV.Finally,the quantum particle swarm algorithm with the dynamic feedback mechanism is proposed to realize optimization of the EV hierarchical scheduling scheme by adaptively adjusting the coefficient of compression-expansion factor.The proposed strategy solves the problem of charging gaps in charging stations during the peak period of EV load by supplementing private shared charging piles.This strategy also achieves the optimal matching of charging facilities and EV load without frequent expansion of charging stations,avoiding safety problems of the power flow caused by centralized EV charging.The proposed algorithm significantly improves efficiency of solving the scheduling strategy while ensuring the accuracy of convergence.(4)The scheduling strategy of EV shared charging sets aside interests of the user side,which causes the actual charging decision of the user to be inconsistent with the grid scheduling plan,resulting in scheduling deviation.For the problem,this paper proposes a multi-objective EV scheduling strategy considering user’s risk preference.Firstly,this strategy proposes a method for judging the risk preference type of EV users based on ensemble clustering.In this case,a variety of clustering methods are used to cluster the risk preference data of EV users to divide the risk preference type of EV users,and then integrate the clustering results based on the concept of integrated learning to further determine the risk preference type of EV users.Secondly,the charging selection criteria for different types of EV user risk preferences are modeled based on behavioral economics,and three charging selection models for EV users,namely,radical,conservative,and balanced,are formulated.Each EV user makes a charging decision based on his own risk preference type,charging demand,battery state of charge,time-of-use price and other factors,which determines whether to accept grid scheduling and selects the corresponding charging mode.Finally,a multi-objective EV scheduling model considering risk preference of users is formulated to benefit EV users while ensuring the stable operation of the distribution network.On the premise of satisfying interests of the user and the grid,the proposed strategy reduces the scheduling deviation caused by the conflict between the EV users’ actual charging decision and the ideal scheduling plan of the grid,so that the effect of actual scheduling tends to be consistent with the effect of ideal scheduling.