| With the current global resources being excessively depleted and air pollution becoming increasingly serious,the development of new energy vehicles in the automotive industry is currently the mainstream trend in the development of the automotive industry.According to domestic and foreign research,as the representative of the new energy vehicles,the bottlenecks in the promotion of electric vehicles mainly include the problem of accurate estimation of the state of charge of electric vehicle battery packs,and the flammable and explosive problems of electric vehicle battery packs.This thesis studies the estimation of the state of charge of the lithium battery pack of electric vehicles,analyzes the influencing factors of the state of charge of the lithium battery through data,and calibrates the state of charge of the actual operating data of the electric vehicle simulated by the lithium iron phosphate battery.Aiming at the nonlinear characteristics between the state of charge of the lithium battery and the voltage and current of the battery,this thesis improves the traditional LSTM network algorithm,and constructs a forward-generating LSTM network model based on time series optimization.The model iteratively utilizes the state information of the lithium battery pack at a fixed time in the past to further estimate the state of charge of the battery at the current time.By comparing the model with the extended Kaman filter algorithm and the BP neural network algorithm,it is concluded that the model has achieved good results in estimating the state of charge of the lithium battery.Finally,this thesis designs and implements an improved electric vehicle lithiumion battery pack measurement and control system.This system integrates the lithium battery pack state-of-charge estimation algorithm,realizes accurate estimation of the lithium battery pack state of charge,and solves existing part defects of the lithium batteries measurement and control system. |
PDF Full Text Request