| Electric vehicle is a means of transportation using clean energy which has energy saving,zero emission and low noise characteristic.With the financial subsidies and policy incentives from governments,electric vehicle inventory will continue to grow rapidly in the future.Electric vehicles are powered by electricity,and their charging load characteristics often depend on the specific travel situation of electric vehicles which is highly flexible and random.Large-scale of electric vehicles being connected to the power grid for charging will increases network loss,decreases power quality and even endangers the operation safety of the grid.Prediction of the charging load of electric vehicles is the foundation and premises for load control,charging infrastructure construction and power grid planning.In this paper,traditional vehicle driving GPS data was obtained to extract and calculate vehicle trip characteristic data.Electric vehicle travel chain simulation model was established by analyzing vehicle’s travel law and putting forward inference process.Based on the travel chain simulation model,the unordered charging load of electric vehicles was accurately predicted under several typical charging scenarios.Specific research contents are as follows:GPS driving data of 400 vehicles in northeast China was collected for more than one year,and the data was filtered according to the integrity and activity of vehicle data.The vehicle traveling data of a typical office worker was selected,and nine travel characteristics variables of vehicle including the trip number,the starting position,the end position,the start time,the end time,the trip mileage,the average speed,the speed variance and the dwell time were calculated.The vehicle trip characteristic data set was built and the vehicle travel law was analyzed.Based on the vehicle travel characteristic data set,the Bayesian Network was used to establish the probability model of the vehicle travel law.And network structure was created considering the physical relationship and correlation between the travel characteristics variable and the network structure obtained by the hill climbing algorithm and the BIC scoring function.The parameters of the Bayesian Network of vehicle travels characteristic were calculated by the vehicle travel data set.Electric vehicle travel chain simulation model was established based on the Bayesian Network.Using the Monte Carlo method and the inference algorithm of Bayesian Networks,a vehicle characteristic variable was inferenced and sampled when some vehicle characteristic variable was given.The travel chain of vehicles in a day was constructed and the travel characteristic data of each trip were calculated.Comparing the simulated data and the original data verifies the accuracy of the model.The object of the study is charging pile dispersed in residential areas which is popularized with high degree.A prototype vehicle was selected and the basic scene of unordered charging of vehicles in the residential area was established.Based on the simulation data of the vehicle’s travel characteristics,the energy consumption,and charging time of each vehicle was estimated,and the vehicle’s unordered charging basic scene was established.The charging load of different scales,different battery states and different charging positions were predicted and analyzed.The results show that the scale of electric vehicles in residential areas is an important factor affecting the charging load of electric vehicles.For each additional electric vehicle in the residential area,the peak value of charging load will increase by 0.08-0.14 times of the charging power of single pile.When the lowest battery SOC level drops,the peak load will drop by 24.2%.When the electric vehicle driver charging at home and workplace,the peak load will drop by 31.8%. |
PDF Full Text Request