Research On Driving Range Prediction And Charging Path Optimization Of Electric Vehicles

Electric vehicles(EVs)have the characteristics of energy saving,environmental protection and high energy utilization efficiency.The promotion of EVs can effectively reduce the dependence on fossil energy in the field of transportation and achieve green,clean and sustainable development.However,the current EV charging time is long,the cruising range is short,and the battery discharge process is easily affected by environmental factors,and users are prone to mileage anxiety.On the other hand,the high charging power of EVs,and the disorderly access of a large number of EVs to the grid for charging will threaten the safe and stable operation of the grid.In addition,the distribution of charging stations in the city is uneven,and it is easy to cause traffic congestion on the roads near the charging station during disorderly charging.Therefore,the mileage of EVs is predicted to determine whether the vehicle can reach the destination,and based on this,combined with the "vehicle-road-grid" information to optimize the charging path of the EVs that need to be charged,it is helpful to improve the user’s experience of EVs.It is of great significance to experience car use,ensure the safe and stable operation of the power grid,and avoid traffic congestion around charging stations.The main research contents of this paper are as follows:(1)An EV-based driving range prediction algorithm based on PCA-XGBoost is proposed.For electric vehicle mileage prediction difficult problems.Firstly,the measured stroke data of EVs were used to process the original data by deleting,filling and sorting methods,and the characteristic parameters reflecting the battery state,motor state and driver driving characteristics of EVs were obtained through feature extraction,so as to obtain good modeling data.Then spearman rank correlation analysis method was used to analyze the correlation between data features.PCA algorithm was used to reduce the dimension of input feature parameters and find the main feature factors that affect the driving distance.Finally,a driving distance prediction model based on PCA-XGBoost was established,and a variety of prediction algorithms were compared.The experimental results verify that the PCA-XGBoost based mileage prediction algorithm has good performance.(2)An EV charging path optimization strategy considering wireless charging and dynamic energy consumption is proposed.To solve the problem of disordered charging of EVs and failing to meet users’ personalized charging needs.Firstly,a charging optimization scheduling model was constructed considering the traffic system,power grid,wireless charging system and dynamic energy consumption of EVs.Then,a charging path optimization algorithm based on Dijkstra and road section weighting was proposed.The optimal charging path can be generated according to the different preferences of EV users.Finally,in order to verify the feasibility and effectiveness of the proposed algorithm,the real road network of Kaifeng Hightech Zone was selected for simulation experiment.Two scenarios with wireless charging system are simulated in MATLAB,and the simulation results are analyzed and compared.Simulation results show that:with the support of wireless charging system,EVs with low initial power can also drive a long distance,which is conducive to providing more travel choices for EV users.At the same time,the algorithm can find the best charging path for EV users with different preferences.The simulation results of 100 EVs show that the algorithm still has good charging path optimization effect with the increase of the number of EVs.