The Drving Range Estimation For Electric Vehicles Based On Battery Status And Driving Energy Consumption

As urban environmental pollution and the scarcity of petroleum resources continue to worsen,electric vehicles are becoming increasingly popular due to their energy-saving and environmentally friendly advantages.Compared to traditional fuel vehicles,electric vehicles have a low energy density,high cost,and short cycle life,leading to limited driving range.The energy consumption rate of vehicles varies greatly under different environmental and driving conditions,resulting in a large estimation error in the driving range,which easily causes "range anxiety" for drivers,especially a series of traffic safety issues caused by unexpected parking due to insufficient driving range.Therefore,an accurate method for estimating driving range has a significant impact on the development of electric vehicles.This paper estimates the driving range of electric vehicles based on vehicle battery state estimation,future driving condition prediction,and vehicle energy consumption calculation.The main research results are as follows:Firstly,five electric vehicle data that had been driving on urban roads for six months were selected,and the characteristics of mileage attenuation and capacity attenuation were constructed.Based on eight features,including real vehicle travel data,a high-precision battery state estimation model was established using the Long Short-Term Memory(LSTM)model.Experimental results show that the average error is 0.57(RMSE)and 0.82%(MAPE).Secondly,to analyze the characteristics of driving conditions,the historical driving data of vehicles are divided into segments.Principal component analysis is utilized to reduce the dimensionality of 13 selected kinematic features,followed by fuzzy C-means clustering to identify three typical driving conditions: low speed,medium speed,and high speed.Based on this,the driving segments are further classified into four subcategories based on driving states.A typical driving condition segment library is constructed using a Markov chain,and Monte Carlo methods are employed to predict future driving conditions.Furthermore,in order to enhance the accuracy of the energy consumption prediction model,a Convolutional Neural Network(CNN)algorithm is applied to classify the predicted driving conditions.Appropriate feature parameters are selected to identify low-speed,medium-speed,and high-speed driving categories.Based on the Light GBM algorithm,energy consumption prediction models are established for each driving condition category,enabling the prediction of energy consumption for different driving conditions.Finally,by combining the available remaining battery capacity estimated by the battery state estimation model and the predicted driving energy consumption based on the future driving condition,the driving range of the electric vehicle is estimated.The results demonstrate that the proposed driving range estimation method achieves good estimation performance with an error of 8.85% compared to real driving conditions.