Research On The Control Method Of Decentralized Charging Pile In Residential Area

For most electric vehicles(EVs),residential area is the most common and ideal charging place,but in practice,EVs are facing the dilemma of "difficult to enter the residential area",especially in some old residential areas with a long construction time and limited capacity of its distribution network.The key to solve the problem is to make a reasonable and orderly charging strategy and control the charging process of EVs.In addition,the way that letting users set their departure time in the existing charging strategies in residential areas may lead to overload.Therefore,this paper studies the orderly charging strategy of EVs in residential areas from two aspects: the prediction of departure time and the orderly charging operation strategy.Based on the theory of long short-term memory(LSTM)network and the analysis of EV's charging mode and user's travel rule,combined with the time series attribute of user's departure time,a LSTM based user departure time prediction model is established.And it realizes the accurate prediction of user's departure time and avoids unbalanced power distribution caused by letting users set the departure time themselves and "cutthroat competition",that is,users can complete charging earlier by further advancing the set departure time.In the aspect of the orderly charging of EVs,with the increase of the permeability of EVs,some of them may not be fully charged,that is,the power at the end of charging does not reach the target power,due to the lower upper limitation of the distribution network in the residential area.This paper proposes a dynamic sequencing mechanism based real-time charging strategy.First,the charging priority parameters are proposed according to the user's charging demand and estimated residence time.The estimated residence time is calculated according to the proposed departure time prediction model.Then the dynamic priority sequence is formed by combining the time monitoring mechanism.Finally,the charging process of EVs are controlled in real time according to the basic load information in the residential area.Considering that there may be EVs that fail to reach the target state of charge(SOC),an evaluation system based on charging completion and total user satisfaction is proposed.The simulation results show that the proposed real-time charging strategy can effectively improve the total user satisfaction,and verify the effectiveness of the proposed prediction model which improves the performance of the real-time charging strategy.In order to solve the problem that the total user satisfaction of the real-time charging strategy declines obviously when the penetration rate of EVs is high,this paper proposes a two-layer optimal charging strategy aiming at the highest total user satisfaction.Based on the assumption that there is no more EVs with charging demand would come after the current time,the power distribution matrix and charging state matrix are successively carried out by using particle swarm optimization algorithm and genetic algorithm.And then control the charging process of EVs according to the optimal solution.The simulation results show that compared with the proposed real-time charging strategy,two-layer optimal charging strategy can further improve the total user satisfaction and meet the charging demands of users.After comparing the two proposed strategies in this paper,the applicable scenarios of the them are analyzed.