Study On Optimal Scheduling Strategy Of Electric Vehicle Clusters In Distribution Power Grid

Electric vehicles(EVs)have the advantages of environmental protection and high efficiency of energy conversion,thus are considered as an important solution for the energy crisis and development of the low-carbon economy,leading the future trend of the automobile industry.The role of EVs in the power grid has two aspects.On the one hand,as a type of largecapacity electrical load,the uncoordinated charging of large-scale EV fleets,which induces high uncertainty,will exert a significant impact on the operation of the power grid.The optimal control strategy of EVs' charging behavior,however,can improve the stability and efficiency of the system operation.On the other hand,as distributed storage units,EVs can be aggregated into distributed energy resources which feedback power to the grid to provide auxiliary services and tackle the intermittency of renewable energy.The EV grid integration being a hot topic for research worldwide,this thesis focuses on the optimization strategy of plugged EVs.The main contents include:We analyze the load characteristics of EVs.Targeting at household EVs,our research focuses on the driving patterns,battery features,and charging modes,and establishes the charging model for EVs.The main structure and management mode of EV charging is introduced.We further analyze the characteristics of centralized optimal control,hierarchical zonal optimal control,and decentralized optimal control,respectively.The multi-agent system is applied to the optimal charging control of EVs in the distribution network.The agents are divided into three levels: the EV agent,regional aggregator agent and distribution network control center agent.A price-based regulation mechanism is used to establish a two-stage optimal dispatch strategy containing the day-ahead dispatch and the quasireal-time adjustment.After that,optimization models of agents at all three levels are developed.The EV agent,seeking to minimize the charging cost and maximize the charging continuity,does the optimization locally.We then transform this optimization problem into a 0-1 mixed integer programming for the convenience of the solution.The regional aggregator agent adjusts the charging load under its jurisdiction using electricity price.With the objective to reduce the peak-valley difference,this agent does a day-ahead optimization and employs genetic algorithm to solve the problem.The distribution network control center agent mainly guarantees secure operation of the distribution network.Taking into account security constraints such as nodal voltage,transformer capacity,and line capacity,this agent issues of quasi-real-time adjustment commands for EV charging when the constraints are violated.Finally,the IEEE 33-bus system is used to show the efficiency of the proposed coordinated control strategy.From the perspective of the EV aggregator,a centralized optimal charging model is established under the scenario of time-of-use electricity price for revenue maximization.This model fully considers constraints such as the state of charge of EVs,vehicle owners' needs for charging and the capacity of distribution transformers.After that,to avoid the problem of "curse of dimensionality" and "communication block" that may occur in the centralized optimization of EV charging under a distribution transformer,we propose a distributed optimal charging control strategy based on Lagrangian relaxation.The capacity constraint of distribution transformers is relaxed and put into the objective function,and the original problem is decomposed into multiple sub-problems for decentralized solutions.Aiming to solve the problem that the Lagrangian relaxation algorithm may oscillate in some periods during the iteration process,a distributed optimal charging control strategy is proposed.Based on the augmented Lagrange method,this control strategy,which improves the convergence and further considers the distribution network constraints,is solved via the alternating direction method of multipliers.Finally,the IEEE 33-bus system is used to show the efficiency of the proposed distributed charging control strategy.When electric vehicles are integrated to microgrids with wind power and photovoltaics,the uncertainty of the wind output,PV output,and electrical load needs to be considered.It is assumed that the microgrid is managed by the EV aggregator.EV batteries are used as a charging load,and as energy storage to optimize the power and energy of the microgrid system as well.The charging and discharging operation modes,as well as the profile of EVs,are adjusted to smooth the fluctuations of wind power,PV,and electrical load.After that,an optimal dispatch model based on day-ahead forecasting data was established to maximize the operating profit of the aggregator.The model takes into account the power balance constraint,power purchasing/selling constraints,and EV charging/discharging constraints,and is converted to a mixed integer linear programming for the solution.Based on this,a model of robust regret optimization for EV charging and discharging control is established using the min-max regret approach,to restrain uncertain disturbance in the system.The model uses a bi-level optimization model,with wind power,PV output and electrical demand as the inner-level decision variables while EV charging/discharging mode and power as the outer-level decision variables.The inner level solves the maximal regret for the feasibility of all the values of the uncertain parameters.The outer level solves the minimum value of the maximal regret.We apply the two-stage Lagrangian relaxation method to solve the problem.Finally,an example is given to compare and analyze the dispatch optimization strategies based on the day-ahead forecasting data and the robust regret,respectively.From the perspective of individual EV coordination,an EV charging management architecture,combining hierarchical control and distributed control,is proposed based on swarm intelligence.Each EV is an adaptive individual,and EVs under one distribution transformer collectively form a charging cluster.The swarm optimization algorithm is used to study the distributed coordinated dispatch of EVs.Considering the state of charge of EVs,vehicle owners' needs for charging and the security constraints,the optimization model of EV coordinated charging based on swarm intelligence is built to minimize the peak-valley difference.In the meantime,the optimization algorithm for EV charging and the algorithm for load fluctuation adjustment based on modified Ant Colony Optimization are proposed and are compared with the Particle Swarm Optimization Algorithm.Finally,simulation is carried out for EV charging optimization under a distribution transformer.