| As the penetration of electric vehicles increases,the number of electric vehicles on motorways increases dramatically,which will further highlight the randomness and aggregation of electric vehicle driving and charging behaviour,thus bringing more severe challenges to the operation of motorway traffic road networks and distribution grids.Therefore,it is important to accurately predict the spatial and temporal distribution of electric vehicle charging loads in the motorway domain.In this paper,the spatio-temporal distribution of EV charging load is predicted quickly and accurately,taking into account various factors such as traffic conditions,with the motorway road network as the background,and its impact on the distribution network is also analysed.The main work is as follows:(1)In order to reasonably predict the spatial and temporal distribution of electric vehicle charging load on highways,a physical information fusion model of " EVs-Traffic-StationDistribution " is studied,and their mutual influence and coupling relationships are analyzed.Firstly,a single electric vehicle model is established: the driving and charging characteristics,travel characteristics and user charging decisions of electric vehicles are physically modeled,and further calibrated using the information data of actual vehicle trips;then the actual road attributes of highways are described and the actual traffic network model is established using graph theory analysis;then,the charging station model is built using the First Come First Service(FCFS)queuing algorithm;finally,a distribution network model is built based on the IEEE30 node topology and node information and line information.(2)To address the problem of the difficulty in predicting the charging load of electric vehicles due to the two-dimensional complexity of their driving and charging activities,a Monte Carlo-based spatio-temporal prediction of the charging load of electric vehicles is studied in conjunction with the traffic flow distribution in the highway domain.Firstly,the spatial and temporal distribution of highway traffic flow is obtained through dynamic simulation of highway traffic flow based on Link Transmission Model(LTM);then the origin-destination(OD)and travel time of electric vehicles are determined by Monte Carlo algorithm,and an OD traffic volume matrix is generated based on the traffic flow data to simulate the trajectory of the EVs on the highway;finally,the spatial-temporal distribution of the electric vehicle load on the highway is obtained by updating the location and charging state of the electric vehicle according to the charging characteristics and charging decisions,and by sampling the simulation with the Monte Carlo algorithm,which verifies the feasibility and effectiveness of the studied method.(3)To address the problems of low efficiency,high memory consumption and difficulty in scaling up practical applications of physical drive methods based on Monte Carlo method in scaled-up EV charging load calculation,a combination of Bi-directional Gated Recurrent Unit(Bi-GRU)and Sequence-to-Sequence(Seq2Seq)algorithms was proposed.An electric vehicle charging load prediction method based on a combination of the Bi-GRU-Seq2 Seq deep learning model and the Monte Carlo method is investigated.The method uses the Bi-GRU-Seq2 Seq deep learning model to effectively capture the travel behaviour information of electric vehicles,encodes their behavioural states as states,establishes the mapping relationship between state codes and charging load sequences through model training,and obtains the prediction results through dynamic semantic vector decoding under the attention mechanism.The algorithm simulation experiments verify that the method can quickly and effectively predict the charging load variation trend of electric vehicles.(4)In order to ensure the safe and stable operation of the distribution network,this paper evaluates the impact on the distribution network before and after the electric vehicle charging load is connected through the calculation and analysis of various evaluation indexes of the distribution network based on the prediction results of the electric vehicle charging load.The IEEE 30-node distribution system is used as the model,and its parameter settings are clarified by combining the node information and branch circuit information,and the distribution system characteristic parameters such as base load,node voltage amplitude and network loss are obtained by using the forward back generation tide calculation method.The simulation shows that as the penetration rate of electric vehicles continues to increase,the load peak-to-valley difference,network loss rate and voltage deviation of the distribution network show a significant increase,especially when the penetration rate of electric vehicles exceeds 30%,which will have an important impact on the safe and stable operation of the distribution network. |
PDF Full Text Request