关键词: 电动汽车, V2G技术, 充放电优化调度, 双层优化模型, 改进粒子群-模拟退火算法

Abstract:

Against the backdrop of carbon peaking and carbon neutrality，electric vehicles（EV）have become increasingly popular as they use cleaner energy，achieve higher efficiency and have made breakthroughs in energy storage.They promise to effectively address the current energy shortage and environmental pollution problems.However，disordered EV charging brings many challenges to the power grid，and greatly affects the safety and reliability of the power grid.Although the traditional time-of-use electricity price strategy has improved the negative impacts of disorderly EV charging，such as the increase of daily load peak-valley difference and the decrease of load rate，it is easy to produce new load peaks and the effect of the current multi-objective optimization strategy is unsatisfactory.The current Vehicle-to-Grid（V2G）technology is able to solve the fundamental problem，requiring a reasonable and efficient EV charging and discharging scheduling strategy.This paper aims to establish a mathematical model of charging and discharging load，improve the particle swarm optimization（PSO）algorithm，and study the orderly charging and discharging optimization strategy and its effect.

First，based on the national household travel survey（NHTS）data，this study deeply analyzes the EV’s driving range，charging start time，charging end time and battery state of charge at the beginning of charging，and establishes the EV charging load model.The EV charging load is simulated and analyzed by Monte Carlo method.The simulation results show that a large number of EVs are connected to the distribution network，and the peak-to-valley difference of the daily load curve increases significantly.The disorderly charging load will have a huge impact on the safe operation of the distribution network.Although the load curve guided by the time-of-use electricity price has a certain peak clipping effect，the effect is not good and a new load peak is generated.

Second，in the case of long parking time and large number of EVs，the PSO method easily gives rise to such problems as local extremum.Considering the characteristics of the EV bi-level optimization model and the advantages and disadvantages of particle swarm and simulated annealing（SA）algorithm，this paper uses the improved PSO-SA hybrid algorithm to solve the above two-layer model.The PSO-SA algorithm addresses the problem when the PSO algorithm falls into the local optimum while the SA algorithm is employed to perturb and optimize the optimal solution obtained by the current PSO，trying to jump out of the local optimal search for a better solution.Our results show the improved PSO-SA achieves a higher efficiency and better optimization accuracy.

Third，this paper proposes a scheduling strategy based on EV bi-level optimization model that fully considers the needs of both the power grid and users.By controlling the charging and discharging power of EVs in different time periods within a day，this paper considers various constraints such as transformer and EV charging and discharging power，and uses the improved PSO-SA algorithm to optimize the first layer model to obtain the daily load curve considering only the demand of the grid side.Taking the first-stage optimization results as constraints，this paper obtains the second-level optimization strategy based on user-side requirements.Taking the minimum charging and discharging cost of the owner as the optimization objective and considering the various constraints of the layer model，this paper obtains the optimization results of the second layer model.The charging and discharging power optimization results of EV in each time period obtained by the second layer model are fed back to the upper layer for the next cycle.The upper and lower models iterate repeatedly until the results meet the termination conditions.

Finally，the optimal solution of EV optimal scheduling based on the two-layer model is obtained.Compared with the load curves before and after optimization，our results show the proposed V2G scheduling strategy based on the bi-level optimization model effectively reduces the load peak-valley differences，increases the load rate and reduces the power costs for EV owners.