Research On Health Assessment And Optimization Technology Of Lithium Battery For Electric Vehicles

As the global energy crisis intensifies,new energy vehicles are receiving increasing attention.Among them,electric vehicles have become the focus of attention in various countries due to their advantages such as low energy consumption and pollution-free emissions.However,as the core component of electric vehicles,power batteries have been restricting the development of electric vehicles because of their low utilization rate of stored energy and short lifespan.Therefore,it is very important for the development of electric vehicles to identify the parameters of the equivalent circuit model and accurately detect the remaining battery capacity and health status.The object of this research is lithium iron phosphate batteries,which are widely used in electric vehicles.Through domestic and foreign research,it has been found that the relevant parameters of batteries have a certain impact on the state of charge(SOC)of batteries,so it is necessary to conduct relevant experiments to analyze the basic characteristics of batteries.Considering the complexity and practicality of the model,the Thevenin model and the secondorder resistance capacitance model are selected as the equivalent circuit models of the battery.The parameters in the Thevenin model are identified offline through the Hybrid Pulse Power Characteristic(HPPC)experiment,and then the recursive least squares method with forgetting factor(FFRLS)is selected to identify the parameters in the second-order resistance capacitance model online.The results show that online identification can still perform stably under relatively complex models,with good adaptability and estimation accuracy.After identifying the model parameters,select 20 identified retired batteries and classify them through Self Organizing Map(SOM).Genetic Algorithm(GA)is used to optimize the BP neural network to estimate the open circuit voltage(OCV)of lithium batteries.A three-layer network system is constructed and a genetic algorithm is introduced to obtain the optimal weight of the BP network.The samples of the neural network are obtained through constant current discharge experiments.After training,the GA-BP neural network accurately predicts the open circuit voltage of the lithium battery based on the input operating voltage and current.The OCV-SOC curve is used to output the remaining power of the battery.Finally,100 sets of test samples are imported for simulation to verify the accuracy of the prediction results.At the same time,a prediction model of battery health(SOH)was established,and the topology and principle of Long Short Term Memory(LSTM)networks were analyzed.On this basis,Particle Swarm Optimization(PSO)algorithm was introduced.The network sample uses battery data disclosed by the NAS APCoE Research Center,and parses the data into time series as input to the LSTM network to obtain prediction results of battery health status.Finally,this paper studies the distributed battery management system(BMS),using the STM32F072RBT6 single chip microcomputer of Italian French company as the main chip,and collecting the voltage,temperature,and other parameters of the battery through the simulation front-end chip SH367306.A GA-BP neural network is used to establish an inputoutput relationship model between operating current(constant),voltage,and battery open circuit voltage.Based on the established model,the SOC of lithium batteries is estimated by hardware.Therefore,from the perspective of safe driving of electric vehicles,battery management systems provide safety assurance for drivers and promote the development of the tram industry.